On-device ai agent-based personal information protection and security inspection method

An on-device AI agent addresses the inefficiencies of conventional methods by providing comprehensive personal information protection and security inspection, ensuring efficient and compliant data handling within user terminals.

KR103046033B1Active Publication Date: 2026-11-10HOME & SERVICE CO LTD
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
KR1020260127080
Authority / Receiving Office
KR · KR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2026-07-10
Publication Date
2026-11-10
Estimated Expiration
2046-07-10

AI Technical Summary

Technical Problem

Conventional personal information protection and security inspection methods are cumbersome for general users, often requiring manual navigation through complex settings and lack context-based judgment, leading to potential data leakage and management inefficiencies, with risks of incomplete or unauthorized data handling.

Method used

An on-device AI agent-based method that comprehensively searches for personal information, calculates risk scores, and generates action scenarios for deletion, masking, or encryption, while verifying the completion of security measures within user-approved authority scopes.

Benefits of technology

Enables efficient, context-aware personal information protection and security inspection within user terminals, minimizing external data transmission and ensuring compliance with organizational policies.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 112026084271925-PAT00004_ABST
    Figure 112026084271925-PAT00004_ABST
Patent Text Reader

Abstract

The present specification describes a method for personal information protection and security inspection based on an on-device AI agent performed by a security inspection device, comprising: a step in which a user authentication unit authenticates a user based on user authentication information and generates user authorization information of the authenticated user; a step in which a personal information search unit sets a personal information search type including personal information image search, PC file search, and internal email search based on the user authorization information and sets search conditions corresponding to the personal information search type; and a step in which the personal information search unit identifies search target data stored in a user terminal or accessible through the user terminal based on the personal information search type and the search conditions, and generates search result information for the search target data. The above search conditions include a search target drive, a search target path, whether subfolders are included, file extension, file name, search keyword, creation time, modification time, last access time, search target mailbox, search period, sender, recipient, whether the email body is searched and whether attachments are searched, and the above search target data includes images, documents, PC files, downloaded files, compressed files, email body, email attachments and browser saved data, and the above search result information may include a file name, save path, file format, file size, creation time, modification time, email subject, attachment file name, email receipt date and whether user-specified keywords are included.
Need to check novelty before this filing date? Find Prior Art

Description

Technology Field

[0001] This specification relates to technology for personal information protection and terminal security inspection, and more specifically, to a method for searching for personal information images, PC files, and internal emails stored on or accessible through a user terminal in an on-device environment, analyzing personal information risk levels, retention periods, contextual information, response scenarios, and PC security setting status using an on-device AI agent, and performing personal information protection measures and security inspection measures within the scope of user approval or delegated authority. Background Technology

[0003] Companies and organizations provide personal information protection guidelines, device security guidelines, and inspection manuals to prevent personal information leakage and maintain a consistent level of security for their members' devices. These security guidelines may include various items such as the storage and deletion of personal information files, management of personal information contained in emails, screen saver settings, network sharing settings, password management, removal of unauthorized programs, and operating system updates.

[0004] However, conventional security check methods typically involved providing users with a manual containing numerous check items and complex configuration procedures, requiring them to refer to the manual to check each item individually. Consequently, users lacking specialized security knowledge often found it difficult to accurately understand the meaning of the check items and may have to spend a significant amount of time locating the corresponding settings within the operating system or external security programs. In particular, items such as screen savers, network sharing, login passwords, installation programs, and operating system updates are managed within separate operating system settings screens or programs, causing inconvenience for users who had to navigate through each setting screen manually.

[0005] In addition, images, documents, downloaded files, emails, and browser usage history containing personal information may exist distributed across various storage locations on user terminals. Conventional users had to individually check each storage location or program to delete or mask personal information, which meant there was a possibility that some personal information might be omitted or unnecessarily stored for a long period.

[0006] Conventional automated security inspection programs may use methods that automatically modify terminal configuration information or collect inspection information from a central server. However, the application of such methods may be restricted depending on user permissions, operating system environments, or organizational security policies, and there is a risk that sensitive information, such as personal information filenames, storage paths, email content, or images, may be transmitted externally.

[0007] Furthermore, conventional personal information detection technologies often rely on structured patterns or keywords to determine whether personal information is included, making it difficult to assess the purpose of retention or whether the retention period has been exceeded based on the context of email bodies, attachment names, folder names, file names, or document content. For example, even for identical name and contact information, the permitted retention period or required measures may vary depending on the business purpose—such as recruitment review, customer service, settlement processing, identity verification, or contract management—but conventional methods have struggled to adequately reflect such context-based judgments.

[0008] Furthermore, even if a user reports that security measures have been completed, there was a problem in verifying whether the relevant settings were actually changed, whether deletion, masking, isolation, or destruction measures regarding personal information were performed, or whether such measures were lawfully executed within the scope of organizational policies and user authority. In addition, if the entire inspection result is stored on a central server for administrators or organizational leaders to review, the concentration of personal information or terminal environment data included in the results on the server may increase the management burden and security risks.

[0009] Therefore, technology is required that guides general users to easily check multiple personal information protection items and PC security settings, and enables personal information retrieval, risk assessment, creation of action scenarios, execution of actions based on user approval or delegation of authority, comparison of states before and after action, and generation of verifiable inspection results in an on-device environment while minimizing the external transmission of original personal information. The problem to be solved

[0011] The purpose of the present specification is to provide a method for personal information protection and security inspection based on an on-device AI agent that comprehensively searches for personal information images, PC files, and internal emails stored on or accessible through a user terminal in an on-device environment, calculates a risk score based on the type, retention period, storage path, contextual information, and action history of the searched personal information, and generates at least one action scenario among deletion, masking, encryption isolation, transfer to a safe folder, safe destruction, or exception handling according to the risk score.

[0012] Furthermore, the purpose of this specification is to provide an on-device AI agent-based personal information protection and security inspection method that compares state information or risk scores before and after action regarding personal information search results, personal information action results, personal information deletion status, and PC security setting status, and determines the level of authority delegation based on the personal information sensitivity of the action target, the recoverability of the action method, user rights, and organizational policy, thereby generating inspection result information that verifies the completion and consistency of security measures performed within the scope of user approval or delegated authority.

[0013] The technical problems that this specification aims to solve are not limited to those mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art to which this specification belongs from the detailed description of the specification below. means of solving the problem

[0015] One aspect of the present specification is a method for personal information protection and security inspection based on an on-device AI agent performed by a security inspection device, comprising: a step in which a user authentication unit authenticates a user based on user authentication information and generates user authorization information of the authenticated user; a step in which a personal information search unit sets a personal information search type including personal information image search, PC file search, and internal email search based on the user authorization information and sets a search condition corresponding to the personal information search type; and a step in which the personal information search unit identifies search target data stored in a user terminal or accessible through the user terminal based on the personal information search type and the search condition, and generates search result information for the search target data. The above search conditions include a search target drive, a search target path, whether subfolders are included, file extension, file name, search keyword, creation time, modification time, last access time, search target mailbox, search period, sender, recipient, whether the email body is searched and whether attachments are searched, and the above search target data includes images, documents, PC files, downloaded files, compressed files, email body, email attachments and browser saved data, and the above search result information may include a file name, save path, file format, file size, creation time, modification time, email subject, attachment file name, email receipt date and whether user-specified keywords are included.

[0016] Additionally, the on-device AI agent management unit performs personal information detection, context analysis, inference of storage purpose, determination of whether the storage period has been exceeded, and calculation of a risk score based on the search result information, and generates a personal information action scenario and an authority delegation level based on the risk score; the personal information action unit performs personal information action based on the personal information action scenario and the authority delegation level and generates personal information action result information; and the inspection result judgment unit generates a first inspection result information corresponding to the personal information search and action process based on the personal information action result information, the risk score before action, and the risk score after action.

[0017] In addition, the personal information deletion and PC security setting management unit acquires the personal information action result information or the first inspection result information, and determines subsequent inspection target items for the personal information deletion status or PC security setting status based on the personal information action result information or the first inspection result information; and the subsequent inspection target items may include the safe folder management status, download folder cleanup status, trash bin cleanup status, browser usage history deletion status, encrypted isolated storage area status, safe destruction completion status, exception handling validity period status, screen saver setting, network sharing setting, operating system password setting, firewall setting, account authority, whether unauthorized programs are installed, and operating system update status.

[0018] Additionally, the personal information deletion and PC security setting management unit may further include the step of acquiring state information prior to action regarding the item subject to subsequent inspection and providing the state information prior to action to the on-device AI agent management unit; the step of the on-device AI agent management unit generating a risk score, a security action scenario, and a delegation level for each inspection item based on the state information prior to action and the first inspection result information; and the step of the personal information deletion and PC security setting management unit performing a personal information deletion measure or a PC security setting measure based on the security action scenario and the delegation level.

[0019] Additionally, the personal information deletion and PC security setting management unit may further include the step of reacquiring post-action status information for the subsequent inspection target item after the personal information deletion measure or the PC security setting measure is performed, and providing the pre-action status information, post-action status information, security measure scenario, authority delegation level, and action execution result to the inspection result judgment unit; and the inspection result judgment unit may further include the step of comparing the pre-action status information and the post-action status information, and generating second inspection result information corresponding to the personal information deletion status and the PC security setting status based on the risk reduction amount of the risk score before the measure and the risk score after the measure.

[0020] Additionally, the inspection result determination unit may further include the step of linking the first inspection result information and the second inspection result information based on the same inspection session identification information, user identification information, terminal identification information, action target identification information, or inspection item identification information, and generating an integrated inspection record that integrates personal information search items, personal information action items, personal information deletion status items, and PC security setting items into a single inspection session unit; the inspection result determination unit may further include the step of calculating the degree of risk reduction and the degree of satisfaction of state change by comparing the information before and after action for each item included in the integrated inspection record; the inspection result determination unit may further include the step of calculating an action consistency score based on the degree of risk reduction, the degree of satisfaction of state change, the degree of authority suitability according to the authority delegation level, and the integrity value of the first inspection result information, the second inspection result information, or the integrated inspection record; and the inspection result determination unit may further include the step of determining the inspection status for each item based on the action consistency score.

[0021] Another aspect of the present specification is a security inspection device for performing personal information protection and security inspection based on an on-device AI agent, comprising: a user authentication unit that authenticates a user based on user authentication information and generates user authorization information of the authenticated user; and a personal information search unit that sets a personal information search type including personal information image search, PC file search, and internal email search based on the user authorization information, sets search conditions corresponding to the personal information search type, identifies search target data stored in a user terminal or accessible through the user terminal based on the personal information search type and the search conditions, and generates search result information for the search target data. The above search conditions include a search target drive, a search target path, whether subfolders are included, file extension, file name, search keyword, creation time, modification time, last access time, search target mailbox, search period, sender, recipient, whether the email body is searched and whether attachments are searched, and the above search target data includes images, documents, PC files, downloaded files, compressed files, email body, email attachments and browser saved data, and the above search result information may include a file name, save path, file format, file size, creation time, modification time, email subject, attachment file name, email receipt date and whether user-specified keywords are included. Effects of the invention

[0023] The present specification provides a method for personal information protection and security checks based on an on-device AI agent that comprehensively searches for personal information images, PC files, and internal emails stored on or accessible through a user terminal in an on-device environment, calculates a risk score based on the type, retention period, storage path, contextual information, and action history of the searched personal information, and generates at least one action scenario among deletion, masking, encryption isolation, transfer to a safe folder, safe destruction, or exception handling according to the risk score.

[0024] In addition, the present specification provides an on-device AI agent-based personal information protection and security inspection method that compares state information or risk scores before and after action regarding personal information search results, personal information action results, personal information deletion status, and PC security setting status, and determines the level of authority delegation based on the personal information sensitivity of the action target, the recoverability of the action method, user rights, and organizational policy, thereby generating inspection result information that verifies the completion and consistency of security measures performed within the scope of user approval or delegated authority.

[0025] The effects obtainable in this specification are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art to which this specification belongs from the description below. Brief explanation of the drawing

[0027] FIG. 1 is a block diagram for illustrating an electronic device related to the present specification. FIG. 2 is a block diagram of an AI device according to one embodiment of the present specification. FIG. 3 is a block diagram illustrating the functional configuration of an on-device AI agent-based personal information protection and security inspection device according to one embodiment of the present specification. FIG. 4 is a flowchart illustrating a method in which a security inspection device according to one embodiment of the present specification searches for personal information using the components of FIG. 3, performs personal information measures based on the analysis results of an on-device AI agent, and generates first inspection result information. FIG. 5 is a flowchart illustrating a method in which a security inspection device according to one embodiment of the present specification checks the personal information deletion status and PC security setting status using the personal information search and action results according to FIG. 4, and generates second inspection result information based on the on-device AI agent analysis results. FIG. 6 is a flowchart illustrating a method for integrating first inspection result information and second inspection result information according to one embodiment of the present specification and verifying inspection results based on an action consistency score. Specific details for implementing the invention

[0028] Hereinafter, embodiments disclosed in this specification will be described in detail with reference to the attached drawings. Identical or similar components regardless of drawing symbols are assigned the same reference number, and redundant descriptions thereof will be omitted. The suffixes "module" and "part" used for components in the following description are assigned or used interchangeably solely for the ease of drafting the specification and do not have distinct meanings or roles in themselves. Furthermore, in describing the embodiments disclosed in this specification, if it is determined that a detailed description of related prior art could obscure the essence of the embodiments disclosed in this specification, such detailed description will be omitted. Additionally, the attached drawings are intended only to facilitate understanding of the embodiments disclosed in this specification; the technical concept disclosed in this specification is not limited by the attached drawings, and it should be understood that they include all modifications, equivalents, and substitutions that fall within the concept and technical scope of this specification.

[0029] Terms including ordinal numbers, such as first, second, etc., may be used to describe various components, but said components are not limited by said terms. These terms are used solely for the purpose of distinguishing one component from another.

[0030] When it is stated that one component is "connected" or "connected" to another component, it should be understood that while it may be directly connected or connected to that other component, there may also be other components in between. On the other hand, when it is stated that one component is "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between.

[0031] Singular expressions include plural expressions unless the context clearly indicates otherwise.

[0032] In this application, terms such as “comprising” or “having” are intended to specify the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0033] FIG. 1 is a block diagram for illustrating an electronic device related to the present specification.

[0034] The above electronic device (100) may include a wireless communication unit (110), an input unit (120), a sensing unit (140), an output unit (150), an interface unit (160), a memory (170), a control unit (180), and a power supply unit (190), etc. Since the components illustrated in FIG. 1 are not essential for implementing the electronic device, the electronic device described herein may have more or fewer components than those listed above.

[0035] More specifically, among the above components, the wireless communication unit (110) may include one or more modules that enable wireless communication between the electronic device (100) and a wireless communication system, between the electronic device (100) and another electronic device (100), or between the electronic device (100) and an external server. Additionally, the wireless communication unit (110) may include one or more modules that connect the electronic device (100) to one or more networks.

[0036] This wireless communication unit (110) may include at least one of a broadcast receiving module (111), a mobile communication module (112), a wireless internet module (113), a short-range communication module (114), and a location information module (115).

[0037] The input unit (120) may include a camera (121) or video input unit for inputting a video signal, a microphone (122) or audio input unit for inputting an audio signal, and a user input unit (123, e.g., a touch key, a mechanical key, etc.) for receiving information from a user. Voice data or image data collected from the input unit (120) may be analyzed and processed into a control command by the user.

[0038] The sensing unit (140) may include one or more sensors for sensing at least one of information within the electronic device, information about the surrounding environment surrounding the electronic device, and user information. For example, the sensing unit (140) may include at least one of a proximity sensor (141), an illumination sensor (142), a touch sensor, an acceleration sensor, a magnetic sensor, a gravity sensor (G-sensor), a gyroscope sensor, a motion sensor, an RGB sensor, an infrared sensor (IR sensor: infrared sensor), a fingerprint sensor (finger scan sensor), an ultrasonic sensor, an optical sensor (e.g., see camera (121)), a microphone (see 122), a battery gauge, an environmental sensor (e.g., a barometer, a hygrometer, a thermometer, a radiation detection sensor, a heat detection sensor, a gas detection sensor, etc.), and a chemical sensor (e.g., an electronic nose, a healthcare sensor, a biometric sensor, etc.). Meanwhile, the electronic device disclosed in this specification can utilize information sensed by at least two of these sensors in combination.

[0039] The output unit (150) is intended to generate output related to sight, hearing, or touch, and may include at least one of a display unit (151), an audio output unit (152), a haptic module (153), and an optical output unit (154). The display unit (151) may form a layered structure with a touch sensor or be formed integrally to implement a touch screen. Such a touch screen functions as a user input unit (123) that provides an input interface between the electronic device (100) and the user, and at the same time can provide an output interface between the electronic device (100) and the user.

[0040] The interface section (160) serves as a passage for various types of external devices connected to the electronic device (100). This interface section (160) may include at least one of a wired / wireless headset port, an external charger port, a wired / wireless data port, a memory card port, a port for connecting a device equipped with an identification module, an audio I / O (Input / Output) port, a video I / O (Input / Output) port, and an earphone port. The electronic device (100) can perform appropriate control related to the connected external device in response to the external device being connected to the interface section (160).

[0041] Additionally, the memory (170) stores data that supports various functions of the electronic device (100). The memory (170) can store a number of application programs (or applications) running on the electronic device (100), data for the operation of the electronic device (100), and commands. At least some of these application programs may be downloaded from an external server via wireless communication. Also, at least some of these application programs may exist on the electronic device (100) from the time of shipment for the basic functions of the electronic device (100) (e.g., phone incoming and outgoing functions, message receiving and outgoing functions). Meanwhile, the application programs may be stored in the memory (170), installed on the electronic device (100), and driven by the control unit (180) to perform the operation (or function) of the electronic device.

[0042] In addition to operations related to the application program, the control unit (180) typically controls the overall operation of the electronic device (100). The control unit (180) can provide or process appropriate information or functions to the user by processing signals, data, information, etc. that are input or output through the components described above, or by running an application program stored in memory (170).

[0043] Additionally, the control unit (180) can control at least some of the components examined together with FIG. 1 in order to run an application program stored in memory (170). Furthermore, the control unit (180) can operate at least two or more of the components included in the electronic device (100) in combination with each other to run the application program.

[0044] The power supply unit (190) receives external power and internal power under the control of the control unit (180) and supplies power to each component included in the electronic device (100). This power supply unit (190) includes a battery, and the battery may be a built-in battery or a replaceable battery.

[0045] At least some of the above components may operate in cooperation with each other to implement the operation, control, or control method of an electronic device according to various embodiments described below. Additionally, the operation, control, or control method of the electronic device may be implemented on the electronic device by running at least one application program stored in the memory (170).

[0046] In this specification, the electronic device (100) may be collectively referred to as a server, and the server may include a cloud server. Additionally, the terminal may include all or part of the configuration of the electronic device (100), and may include a tablet PC.

[0047] FIG. 2 is a block diagram of an AI device according to one embodiment of the present specification.

[0048] The AI ​​device (20) may include an electronic device including an AI module capable of performing AI processing, or a terminal including the AI ​​module. Additionally, the AI ​​device (20) may be configured to be included as at least a part of the configuration of the electronic device (100) shown in FIG. 1 to perform at least a part of the AI ​​processing together.

[0049] The above AI device (20) may include an AI processor (21), memory (25) and / or a communication unit (27).

[0050] The above AI device (20) is a computing device capable of learning a neural network and can be implemented as various electronic devices such as a terminal, desktop PC, laptop PC, tablet PC, etc.

[0051] The AI ​​processor (21) may be configured to use an artificial intelligence model stored in memory (25) to detect or classify personal information contained in images, documents, files, or mail data obtained from a user terminal, and to generate risk scores, action scenarios, or security check result information regarding said personal information.

[0052] Specifically, the AI ​​processor (21) receives image data, document data, file metadata, or mail data searched by an image search unit, a PC file search unit, or an internal mail search unit, and can determine whether the data contains personal information.

[0053] The above personal information may include at least one of a person's face, identification card, resident registration number, driver's license number, passport number, name, address, contact information, signature, account information, biometric information, employee number, customer identification information, contract identification information, or document identification information.

[0054] The AI ​​processor (21) can detect personal information areas or personal information strings included in images, documents, or mail data using at least one of an object detection model, an image classification model, an optical character recognition model, an image segmentation model, a document layout analysis model, or a natural language processing model.

[0055] For example, the AI ​​processor (21) can detect an ID card object, a driver's license object, a passport object, a face area, or a signature area in an image, and identify a character area or biometric identification area included in the object as a personal information area.

[0056] Additionally, the AI ​​processor (21) can extract characters from an image or document using an optical character recognition model and determine the type of personal information based on the format, location, surrounding characters, or context of the extracted string.

[0057] For example, the AI ​​processor (21) can determine whether the extracted string has a pattern corresponding to a resident registration number, phone number, account number, address, email address, employee number, or customer identification number.

[0058] The AI ​​processor (21) can analyze contextual information including at least one of an email subject, email body, attachment file name, attachment content, file name, folder name, document body, creation time, modification time, reception time, and sender information.

[0059] The AI ​​processor (21) can infer the purpose of storage of data containing personal information based on the contextual information. The purpose of storage may include at least one of recruitment review, customer response, settlement processing, identity verification, contract management, delivery processing, complaint processing, internal audit, or security check.

[0060] The AI ​​processor (21) can determine the allowed retention period for the image, document, file, or mail data by comparing the inferred retention purpose with the preset retention period policy.

[0061] The AI ​​processor (21) can determine whether the storage period has been exceeded by comparing at least one of the data creation time, modification time, reception time, last access time, or user input time with the allowed storage period.

[0062] The AI ​​processor (21) can calculate a risk score for each data based on at least one of the following: personal information type, personal information sensitivity, whether the retention period has been exceeded, the risk level of the storage path, whether encryption is enabled, the possibility of external sharing, the history of non-action, the history of exception handling, or the history of external transmission and reception.

[0063] For example, the AI ​​processor (21) may assign a relatively high personal information sensitivity score to files containing identification cards, resident registration numbers, biometric information, or account information, and may assign a relatively high storage path risk score to files stored in a download folder, temporary folder, shared folder, or external synchronization folder.

[0064] The AI ​​processor (21) can generate an action scenario for the data when the calculated risk score is greater than or equal to a preset standard. The action scenario may include at least one of deletion, AI automatic masking, manual masking, encryption isolation, transfer to a safe folder, safe destruction, exception handling, request for administrator approval, or request for user confirmation.

[0065] The AI ​​processor (21) can determine the level of authority delegation of the action scenario based on at least one of the personal information sensitivity of the data to be taken, the recoverability of the action method, user rights, organizational policy, whether exception processing is possible and whether administrator approval is required.

[0066] The above authority delegation level can be determined as any one of immediate execution, user verification required, explicit approval required, prior delegation possible, or administrator approval required.

[0067] The AI ​​processor (21) can provide the determined authority delegation level to the personal information action unit or the inspection result judgment unit, and the personal information action unit can execute an action scenario or request user approval or administrator approval according to the authority delegation level.

[0068] The AI ​​processor (21) can generate location information of a personal information area included in an image or document image and provide the location information to a personal information processing unit or a masking processing unit.

[0069] The personal information processing unit can apply blur processing, mosaic processing, color replacement, character removal, area deletion, patch insertion, or other de-identification processing to the area based on the location information of the personal information area provided by the AI ​​processor (21).

[0070] The AI ​​processor (21) can calculate the detection reliability of each personal information area and can determine the personal information area with a detection reliability greater than or equal to a preset standard as an automatic masking target.

[0071] If the detection reliability is below a preset standard, the AI ​​processor (21) can determine the area as a target for user verification and display candidate areas that may be personal information on the user interface.

[0072] The user can approve the privacy area detected by the AI ​​processor (21), modify the size or location of the privacy area, or add a new masking area.

[0073] In one embodiment, the AI ​​processor (21) can calculate the possibility of re-detecting personal information for an image or document image after masking processing.

[0074] The AI ​​processor (21) can reapply at least one of an optical character recognition model, an object detection model, an image classification model, or a document layout analysis model to the image after masking processing, and determine whether personal information strings, face areas, ID card areas, or document identification information are re-detected in the masked area.

[0075] The AI ​​processor (21) determines that the masking is complete when the possibility of re-detection is below a preset threshold, and when the possibility of re-detection exceeds the threshold, it can generate additional masking candidates with changed masking intensity, masking area size, masking pattern, or masking method.

[0076] The AI ​​processor (21) generates feature information for each of the original image and the comparison target image, and can determine whether the comparison target image is identical or similar to the original image based on the similarity between the feature information.

[0077] For example, even if the size, resolution, compression ratio, rotation direction, file name, storage path, or some area of ​​the image is changed, the AI ​​processor (21) can determine whether it is substantially the same personal information image by using image feature vectors or perceptual feature information.

[0078] The AI ​​processor (21) may provide the results of the identity or similarity determination to the personal information deletion management unit, the inspection result determination unit, or the report generation unit. Based on the above determination results, the personal information deletion management unit may apply the scheduled time of destruction of the existing file to the file, or manage the same or similar file as a target for destruction.

[0079] In one embodiment, the AI ​​processor (21) can analyze the OS security setting status of the user terminal or the external transmission and reception activity status.

[0080] The AI ​​processor (21) can analyze at least one of screen saver settings, network sharing settings, operating system password settings, firewall settings, account permissions, whether unauthorized programs are installed, or operating system update status, and can identify vulnerable items based on the analysis results.

[0081] The AI ​​processor (21) can generate at least one security action scenario for each vulnerable item, such as calling a settings screen, direct user action, one-click action, automatic setting change, request for exception approval, or administrator reporting.

[0082] Additionally, the AI ​​processor (21) can detect external transmission and reception events that may leak personal information by analyzing at least one of browser connection information, network connection information, cloud storage connection information, file upload event, mail attachment transmission event, or messenger file transmission event.

[0083] The AI ​​processor (21) can calculate the risk of leakage based on the personal information risk score of the file to be transmitted and the reliability of the site to be transmitted, and can generate at least one action scenario among real-time warning, transmission suspension, encryption recommendation, masking recommendation, or administrator approval request according to the risk of leakage.

[0084] In one embodiment, the AI ​​processor (21) can calculate the amount of risk reduction by comparing the first risk score before the execution of the action scenario and the second risk score after the execution of the action scenario.

[0085] The AI ​​processor (21) may determine that the measure is completed if the amount of risk reduction is greater than or equal to the standard for each type of measure, and may determine that the measure is partially taken, requires further verification, or failed if the amount of risk reduction is less than the standard for each type of measure.

[0086] The AI ​​processor (21) can provide at least one of the following to the inspection result judgment unit or report generation unit: personal information detection result, personal information type, contextual information, storage purpose, allowed storage period, risk score, action scenario, authority delegation level, risk score before and after action, risk reduction amount, possibility of masking re-detection, and action completion judgment result.

[0087] Accordingly, the AI ​​processor (21) can perform at least one of the following: detection of personal information contained in images, documents, files or mail data, identification of personal information areas, classification of personal information types, context-based inference of storage purpose, determination of allowed storage period, calculation of risk score, generation of action scenarios, determination of authority delegation level, verification of masking re-identification possibility, calculation of leakage risk of external transmission and reception events, and determination of completion based on risk reduction amount before and after action.

[0088] Meanwhile, the AI ​​processor (21) that performs the functions described above may be a general-purpose processor (e.g., CPU), but may be an AI-dedicated processor for artificial intelligence learning (e.g., GPU, graphics processing unit).

[0089] The memory (25) can store various programs and data required for the operation of the AI ​​device (20). The memory (25) can be implemented as non-volatile memory, volatile memory, flash memory, hard disk drive (HDD), or solid-state drive (SDD). The memory (25) is accessed by the AI ​​processor (21), and the AI ​​processor (21) can perform reading / writing / modification / deletion / updating of data. Additionally, the memory (25) can store a neural network model (e.g., a deep learning model) generated through a learning algorithm for data classification / recognition according to one embodiment of the present specification.

[0090] Meanwhile, the AI ​​processor (21) may include a data learning unit that learns a neural network for data classification / recognition. For example, the data learning unit may learn a deep learning model by acquiring training data to be used for learning and applying the acquired training data to a deep learning model.

[0091] The communication unit (27) can transmit the AI ​​processing results by the AI ​​processor (21) to an external electronic device.

[0092] Here, external electronic devices may include other terminals and / or servers (cloud servers).

[0093] Meanwhile, although the AI ​​device (20) illustrated in FIG. 2 is described by functionally separating it into an AI processor (21), memory (25), and communication unit (27), the aforementioned components may be integrated into a single module and referred to as an AI module or an artificial intelligence (AI) model.

[0094] FIG. 3 is a block diagram illustrating the functional configuration of an on-device AI agent-based personal information protection and security inspection device according to one embodiment of the present specification.

[0095] Referring to FIG. 3, the security inspection device (1000) may include a user authentication unit (1100), a personal information search unit (1200), a personal information action unit (1300), a personal information deletion and PC security setting management unit (1400), an on-device AI agent management unit (1500), an inspection result determination unit (1600), a report generation unit (1700), and a report verification management unit (1800).

[0096] security inspection device (1000) It can be implemented as an application, program, security inspection agent, or on-device AI agent installed and running on a user terminal. The security inspection device (1000) can be executed by a processor of the user terminal and can perform at least some processing necessary for searching for personal information, detecting personal information, taking measures against personal information, checking PC security settings, calculating risk scores, generating action scenarios, and verifying inspection results within the user terminal.

[0097] Accordingly, original data of images, documents, files, or emails containing personal information may be searched, analyzed, or processed within the user terminal without being transmitted to an external server. However, depending on the embodiment, at least some of user authentication information, inspection policies, security policy versions, license information, statistical information by organization, or summary information of inspection results may be transmitted to and received with an external server.

[0098] User authentication unit (1100)The device receives user authentication information from the user and may allow access to the security inspection device (1000) based on the entered user authentication information. The user authentication information may include at least one of a user type, name, employee number, affiliation information, job information, terminal identification information, and an authentication token.

[0099] The user type may include at least one of a general user, a security officer, an administrator, a report approver, or an agent user. The user authentication unit (1100) may set the range of functions provided by the security inspection device (1000) differently depending on the user type or authorization information. For example, general users may be provided with functions for searching personal information, taking measures on personal information, checking PC security settings, and generating reports, while report approvers may be additionally provided with functions for checking and approving inspection reports from other users.

[0100] Additionally, the user authentication unit (1100) may provide user authorization information to the on-device AI agent management unit (1500) or the inspection result determination unit (1600). The on-device AI agent management unit (1500) may use the user authorization information to determine whether a specific action scenario can be executed immediately, requires user verification, or requires administrator approval.

[0101] Personal Information Search Department (1200) It can search for personal information related data that is stored on a user terminal or accessible through a user terminal. The personal information search unit (1200) can set at least one of images, documents, PC files, downloaded files, attachments, internal emails, email bodies, email attachments, and browser stored data as a search target.

[0102] The personal information search unit (1200) may receive search conditions from a user, including at least one of a search target drive, a search target path, a file extension, a filename, a search keyword, a search target mailbox, a search period, whether to search for attachments, and whether to include subfolders. The personal information search unit (1200) may identify search target data based on the entered search conditions and generate a list of the identified search target data.

[0103] In one embodiment, the personal information search unit (1200) can generate an image search result for an image file that includes at least one of a thumbnail, a file name, a storage path, a file format, a file size, a creation time, and a modification time. Additionally, the personal information search unit (1200) can generate a file search result for a PC file that includes at least one of a file name, a storage path, an extension, a document body, metadata, and whether user-defined keywords are included.

[0104] When an internal email is set as the search target, the personal information search unit (1200) connects to an email program or groupware installed on a user terminal and can obtain a search result for at least one of an email subject, email body, attachment file name, attachment content, sender information, recipient information, and time of receipt. The personal information search unit (1200) can provide the search result to the on-device AI agent management unit (1500) or the personal information action unit (1300).

[0105] Personal Information Protection Department (1300) It receives user action input regarding images, documents, files, or mail searched by the personal information search unit (1200) and can perform personal information protection measures according to the selected action method. The action method performed by the personal information action unit (1300) may include at least one of deletion, AI automatic masking, manual masking, encryption isolation, safe folder transfer, safe destruction, exception handling, and administrator approval request.

[0106] The personal information action unit (1300) provides search results in the form of a list, table, thumbnail, or preview and can receive a selection of the action target from the user. The user can select all search results or select some search results individually to perform the action.

[0107] If deletion is selected, the personal information handling unit (1300) may delete the selected image, file, or email or move it to the trash. According to an embodiment, the personal information handling unit (1300) may maintain the deleted data in a temporary storage area until the current search session ends, and restore the deleted data to its original location upon the user's restore input.

[0108] When AI automatic masking is selected, the personal information action unit (1300) receives location information, personal information type, and detection confidence of a personal information area from the on-device AI agent management unit (1500) or the AI ​​device, and may apply blur processing, mosaic processing, color replacement, character removal, area deletion, patch insertion, or other de-identification processing to the area. Areas with a detection confidence level below a preset standard may be marked as a target for user verification, and the user may approve the area, modify the size or location of the masking area, or add a new masking area.

[0109] If encryption isolation is selected, the personal information action unit (1300) may encrypt the selected file and move it to an isolated storage area or safe folder separated from the general storage area. If safe destruction is selected, the personal information action unit (1300) may perform deletion processing for the selected file to restrict recovery. If exception processing is selected, the personal information action unit (1300) may set the data to be excluded from deletion, masking, isolation, or destruction, and may store at least one of the exception reason, exception validity period, exception processing user, and exception processing time.

[0110] Personal Information Deletion and PC Security Settings Management Department (1400) It can check the personal information deletion status and the PC security setting status of the user terminal, and provide a path for action corresponding to items requiring action. The personal information deletion status may include at least one of a safe folder management status, a download folder cleanup status, a trash bin cleanup status, and a browser usage history deletion status.

[0111] The personal information deletion and PC security setting management unit (1400) can check at least one of whether a safe folder is created, the retention period of files stored in the safe folder, the scheduled time of destruction, the remaining retention period, and whether they are subject to destruction. In addition, it can check at least one of the number of files stored in the download folder, whether files stored in the trash bin exist, whether the browser usage history deletion function can be called, and the deletion items per browser.

[0112] The PC security setting status may include at least one of screen saver settings, network sharing settings, operating system password settings, firewall settings, account permissions, whether unauthorized programs are installed, operating system update status, and the results of an external security program check. The personal information deletion and PC security setting management unit (1400) can obtain current status information of the check items and identify vulnerable items by comparing them with a pre-configured check policy or organizational security policy.

[0113] For example, the personal information deletion and PC security setting management unit (1400) can check whether the screen saver waiting time is less than or equal to the standard time, whether the logon screen is set to be displayed when the screen saver is released, whether the sharing function for a private network or a public network is enabled, whether an unauthorized program is installed, and whether the operating system update is not performed.

[0114] The personal information deletion and PC security setting management unit (1400) may call at least one of an operating system setting screen, a program removal screen, an operating system update screen, a browser history deletion screen, an external security program execution screen, or a personal information management program execution screen for items that do not meet the inspection criteria. According to an embodiment, the personal information deletion and PC security setting management unit (1400) may directly change some setting values ​​within the scope of user approval or delegated authority.

[0115] On-device AI agent management department (1500) It can analyze at least one of personal information search results, file metadata, mail context information, image analysis results, personal information deletion status, PC security setting status, and external transmission / reception activities by using a lightweight artificial intelligence model running on a user terminal or a dedicated edge environment connected to a user terminal.

[0116] The on-device AI agent management unit (1500) can calculate a risk score for each data based on at least one of the following: personal information type, personal information sensitivity, whether the retention period has been exceeded, the risk level of the storage path, whether encryption is enabled, the possibility of external sharing, the history of non-action, the history of exception handling, or the history of external transmission and reception. For example, a relatively high personal information sensitivity score may be assigned to a file containing an identification card, resident registration number, biometric information, or account information, and a relatively high storage path risk score may be assigned to a file stored in a download folder, temporary folder, shared folder, or external synchronization folder.

[0117] Additionally, the on-device AI agent management unit (1500) can analyze contextual information including at least one of an email subject, email body, attachment file name, attachment content, file name, folder name, document body, creation time, modification time, reception time, and sender information, and infer the purpose of storage of data containing personal information. The purpose of storage may include at least one of recruitment review, customer response, settlement processing, identity verification, contract management, delivery processing, complaint processing, internal audit, or security check. The on-device AI agent management unit (1500) can compare the inferred purpose of storage with a pre-set storage period policy to determine the allowed storage period and whether the storage period has been exceeded.

[0118] The on-device AI agent management unit (1500) can generate action scenarios based on the calculated risk score or whether the retention period has been exceeded. The action scenarios may include at least one of deletion, AI automatic masking, manual masking, encrypted isolation, transfer to a safe folder, safe destruction, exception handling, user confirmation request, administrator approval request, calling a settings screen, one-click action, or automatic setting change.

[0119] The on-device AI agent management unit (1500) can determine the authority delegation level of an action scenario based on at least one of the personal information sensitivity of the action target, the recoverability of the action method, user rights, organizational policy, whether exception processing is possible, whether administrator approval is required, and the scope of prior delegation. The authority delegation level can be determined as any one of immediate execution, user verification required, explicit approval required, prior delegation possible, or administrator approval required.

[0120] In one embodiment, the on-device AI agent management unit (1500) can detect a personal information area included in an image or document image and calculate the possibility of re-detection of personal information for the image after masking processing. If the possibility of re-detection exceeds a preset standard, the on-device AI agent management unit (1500) can generate additional masking candidates by changing the masking intensity, the size of the masked area, the masking pattern, or the masking method.

[0121] In one embodiment, the on-device AI agent management unit (1500) can detect external transmission and reception events that may leak personal information by analyzing at least one of browser connection information, network connection information, cloud storage connection information, file upload event, mail attachment transmission event, or messenger file transmission event. The on-device AI agent management unit (1500) can calculate the risk of leakage based on the personal information risk score of the file to be transmitted and the reliability of the site to be transmitted, and can generate at least one action scenario among real-time warning, transmission suspension, encryption recommendation, masking recommendation, or administrator approval request according to the risk of leakage.

[0122] Inspection result judgment unit (1600) It can obtain at least one of status information, risk score, action scenario, authority delegation level, and action execution result for a plurality of inspection items from the personal information search unit (1200), personal information action unit (1300), personal information deletion and PC security setting management unit (1400), and on-device AI agent management unit (1500).

[0123] The inspection result determination unit (1600) can compare the state information before action and the state information after action regarding the personal information search results, personal information action results, personal information deletion status and PC security setting status. For example, the inspection result determination unit (1600) can compare the number of unhandled personal information items, the number of files in the download folder, the number of files in the Recycle Bin, the screen saver setting status, whether network sharing is enabled, whether unauthorized programs exist, or the operating system update status before and after action.

[0124] Additionally, the inspection result judgment unit (1600) can calculate the amount of risk reduction by comparing the first risk score before the execution of the action scenario with the second risk score after the execution of the action scenario. If the amount of risk reduction is greater than or equal to the standard for each type of action, the inspection result judgment unit (1600) can determine the corresponding action as completed, and if the amount of risk reduction is less than the standard for each type of action, it can determine the corresponding action as partial action, requiring additional verification, or failed action.

[0125] The inspection result judgment unit (1600) can determine the status of each inspection item as insufficient, caution, excellent, completed, partial action, additional verification required, action failed, or conditionally completed. The inspection result judgment unit (1600) can also generate at least one of the following: the reason for judgment of each inspection item, current status, target status, recommended action method, action scenario, authority delegation level, risk score before action, risk score after action, and risk reduction amount.

[0126] The inspection result judgment unit (1600) can generate first inspection result information by linking the search execution information, search result information, action execution information, and action completion judgment result generated during the personal information search and action process with the inspection session identification information. Additionally, the inspection result judgment unit (1600) can generate second inspection result information, including state information before and after action regarding the personal information deletion status and PC security setting status, the result of executing the action scenario, and the amount of risk reduction.

[0127] Report generation unit (1700) The inspection report can be generated based on the first inspection result information and the second inspection result information generated by the inspection result judgment unit (1600). The inspection report may include at least one of the following: status by item, risk score by item, risk reduction amount before and after action, whether essential inspection items are satisfied, overall score, overall grade, priority action items, result of action scenario execution, and authority delegation level.

[0128] The report generation unit (1700) can determine items recommended for priority action based on risk level, necessity, degree of non-action, degree of exceeding the retention period, or possibility of external leakage among items judged to be insufficient, cautionary, partial action, or requiring additional verification. Additionally, the report generation unit (1700) can generate natural language security insights to be provided to a user or administrator based on the analysis results of the on-device AI agent management unit (1500).

[0129] The above security insights may include at least one of the following: the amount of personal information stored per user, the number of unaddressed items per organization, recurring vulnerability items, items exceeding the retention period, external transmission / reception events, the action completion rate, major risk causes, and recommended improvement measures. The report generation unit (1700) may display the detailed judgment content of the inspection item, the basis for calculating the risk score, and the improvement method when the user selects the status of the inspection item.

[0130] Report Verification Management Department (1800) It can generate external verification information corresponding to the inspection report generated by the report generation unit (1700) and support the external verifier's verification and approval of the report. The external verification information may include at least one of user identification information or non-identification information, inspection status by item, overall score, overall grade, priority action item, time of action completion, policy version, authority delegation level, risk reduction amount, and integrity verification information.

[0131] The report verification management unit (1800) can separate the original inspection information obtained during the personal information search process from the result information required for external verification. The original inspection information may include a filename, a storage path, an email subject, an email body, an original image, or actual personal information values. The result information required for external verification may include at least one of the following: status by item, whether the action has been completed, a risk score, a risk reduction amount, an overall grade, a time of completion of the inspection, a policy version, and integrity verification information.

[0132] The report verification management unit (1800) can exclude original inspection information from external verification information and generate report delivery information using only result information required for external verification. The report delivery information can be generated in the form of a copyable string, an optically readable code such as a QR code, link information, or a verification token.

[0133] The report verification management unit (1800) may include a creation time, an expiration time, and integrity verification information in the report delivery information. The integrity verification information may include at least one of a hash value, a message authentication code, a digital signature, encryption information, or the root hash value of a hash tree. The report verification management unit (1800) verifies the validity period and integrity of the report delivery information and may allow the display or approval of the inspection result only if the verification is successful.

[0134] The report verification management unit (1800) may provide a user interface that allows an external verifier to perform approval or rejection input after verifying the inspection results. The report verification management unit (1800) may generate approval information including at least one of approver information, approval time, approval or rejection status, and reason for rejection.

[0135] Accordingly, the security inspection device (1000) can provide a series of processes through a single on-device interface, including searching for personal information images, PC files and internal emails, taking measures on personal information, checking the status of personal information deletion, checking PC security settings, calculating a risk score, generating a measure scenario, determining the level of authority delegation, determining completion based on the amount of risk reduction before and after the measure, generating a report, and obtaining approval from an external verifier.

[0136] In addition, the security inspection device (1000) can restrict the external transmission of the original personal information or detailed search information, while also supporting an external verifier to check the user's inspection results, whether the action has been completed, whether the action has been taken within the scope of delegated authority, and whether the risk has been reduced before and after the action.

[0137] FIG. 4 is a flowchart illustrating a method in which a security inspection device according to one embodiment of the present specification searches for personal information using the components of FIG. 3, performs personal information measures based on the analysis results of an on-device AI agent, and generates first inspection result information.

[0138] Referring to Fig. 4, first The user authentication unit (1100) of the security inspection device (1000) can authenticate a user based on user authentication information (S4100).

[0139] User authentication information may include at least one of the user's name, employee number, affiliation information, job information, user type, terminal identification information, and authentication token. The user type may include at least one of a general user, security officer, administrator, report approver, or agent user.

[0140] The user authentication unit (1100) can provide the user type or user authority information of the authenticated user to the personal information search unit (1200), the personal information deletion and PC security setting management unit (1400), and the on-device AI agent management unit (1500). Accordingly, the security inspection device (1000) can determine at least one of the personal information search range, the personal information actionable range, whether automatic action is possible, whether an administrator approval request is possible, and whether a report can be generated, depending on the user authority.

[0141] Next, The personal information search unit (1200) can set personal information search types and search conditions based on user authority information (S4200).

[0142] The personal information search type may include at least one of personal information image search, PC file search, and internal email search. Personal information image search may be a type for searching for data that may contain personal information in image files or document images stored on a user terminal. PC file search may be a type for searching for data that may contain personal information in document files, spreadsheet files, compressed files, audio recording files, downloaded files, or other electronic files stored on a user terminal. Internal email search may be a type for searching for data that may contain personal information in at least one of the email body, email subject, attachment filename, and attachment content stored in an email program or groupware installed on a user terminal.

[0143] The search conditions may include at least one of the following: a search target drive, a search target path, whether to include subfolders, a file extension, a filename, a search keyword, a creation time, a modification time, a last access time, a search target mailbox, a search period, a sender, a recipient, whether to search the email body, and whether to search for attachments.

[0144] Next, The personal information search unit (1200) can identify search target data stored in the user terminal or accessible through the user terminal based on the set personal information search type and search conditions, and generate search result information (S4300).

[0145] The data to be searched may include at least one of images, documents, PC files, downloaded files, compressed files, email bodies, email attachments, or browser saved data.

[0146] The personal information search unit (1200) can generate search execution information and search result information for the data to be searched. The search execution information may include at least one of search session identification information, search type, search target range, search condition, search start time, search end time, and identification information of the search execution module. The search result information may include at least one of file name, storage path, file format, file size, creation time, modification time, email subject, attachment file name, email receipt date, and whether user-specified keywords are included.

[0147] The personal information search unit (1200) can provide the generated search result information to the on-device AI agent management unit (1500). Additionally, the personal information search unit (1200) can provide a list of data among the search results that may be subject to user action to the personal information action unit (1300).

[0148] Next, The on-device AI agent management unit (1500) can perform personal information detection, context analysis, and risk score calculation based on search result information provided by the personal information search unit (1200) (S4400).

[0149] The on-device AI agent management unit (1500) can perform at least one of rule-based personal information pattern detection, optical character recognition, object detection, document layout analysis, natural language processing, or on-device AI model analysis on images, documents, files, or mail data.

[0150] In the case of image data, the on-device AI agent management unit (1500) can detect personal information areas corresponding to identification cards, faces, resident registration numbers, driver's license numbers, passport numbers, names, addresses, contact information, account information, signatures, or document identification information using at least one of an object detection model, an image classification model, an optical character recognition model, or an image segmentation model.

[0151] In the case of PC files, the on-device AI agent management unit (1500) can identify files that may contain personal information based on at least one of the document body, file name, file metadata, or custom keywords. In the case of internal emails, the on-device AI agent management unit (1500) can identify emails that may contain personal information by analyzing at least one of the email subject, email body, attachment file name, attachment content, sender information, recipient information, and email reception date.

[0152] Additionally, the on-device AI agent management unit (1500) can determine whether the purpose of storage and the storage period of personal information have been exceeded based on the contextual information of the data to be searched. The contextual information may include at least one of an email subject, email body, attachment file name, attachment content, file name, folder name, document body, creation time, modification time, reception time, and sender information.

[0153] The on-device AI agent management unit (1500) can analyze contextual information to infer the purpose of storage of data containing personal information. The purpose of storage may include at least one of recruitment review, customer response, settlement processing, identity verification, contract management, delivery processing, complaint processing, internal audit, or security check. The on-device AI agent management unit (1500) can determine the allowable storage period of the data by comparing the inferred purpose of storage with a pre-set storage period policy, and can determine whether the storage period has been exceeded by comparing the data's creation time, modification time, reception time, or last access time with the allowable storage period.

[0154] The on-device AI agent management unit (1500) can calculate a risk score for each search target data based on at least one of the following: personal information type, personal information sensitivity, number of personal information detections, personal information detection reliability, whether the retention period has been exceeded, risk of the storage path, whether encryption is enabled, possibility of external sharing, history of non-action, history of exception handling, and history of external transmission and reception.

[0155] Next, The on-device AI agent management unit (1500) can generate action scenarios and authority delegation levels based on risk scores (S4500).

[0156] Action scenarios may include at least one of deletion, AI automatic masking, manual masking, encrypted isolation, safe folder transfer, safe destruction, exception handling, user confirmation request, administrator approval request, or additional review request.

[0157] For example, for a file whose risk score is above the first criterion and has exceeded the allowed retention period, the on-device AI agent management unit (1500) may generate a deletion or safe destruction scenario. For a file whose risk score is above the second criterion and requires business retention, the on-device AI agent management unit (1500) may generate an encrypted isolation or safe folder transfer scenario. If a personal information area within an image is detected, the on-device AI agent management unit (1500) may generate an AI automatic masking or manual masking scenario.

[0158] Additionally, the on-device AI agent management unit (1500) can determine the level of authority delegation of the action scenario based on at least one of the personal information sensitivity of the action target, the recoverability of the action method, user rights, organizational policy, whether exception processing is possible, and whether administrator approval is required.

[0159] The authority delegation level can be determined as any one of the following: immediately executable, user verification required, explicit approval required, pre-delegation possible, or administrator approval required. For example, calling a preview screen or providing a configuration path can be classified as immediately executable, file encryption isolation can be classified as requiring user verification, and secure destruction or high-risk email deletion can be classified as requiring explicit approval or administrator approval.

[0160] The on-device AI agent management unit (1500) can provide at least one of the generated action scenario, authority delegation level, location information of the personal information area, type of personal information, detection reliability, risk score, and whether the retention period has been exceeded to the personal information action unit (1300).

[0161] Next, The personal information action unit (1300) provides personal information search results and action candidates to the user based on the list of action targets provided by the personal information search unit (1200) and the action scenarios and authority delegation levels provided by the on-device AI agent management unit (1500), and can perform personal information actions according to user input or authority delegation levels (S4600).

[0162] The personal information search results may include at least one of a list of searched files or emails, filenames, storage paths, file formats, file sizes, creation dates, modification dates, email subject lines, attachment filenames, personal information detection types, detection reliability, storage purposes, allowed retention periods, whether the retention period has been exceeded, risk scores, and current action status.

[0163] The personal information action unit (1300) may select at least one action method among deletion, AI automatic masking, manual masking, encryption isolation, transfer to a safe folder, safe destruction, or exception handling for all or part of the search results.

[0164] The personal information action unit (1300) can control the method of executing actions differently depending on the level of authorization delegation. In the case of an action that can be executed immediately, the personal information action unit (1300) can call the action path without separate approval. In the case of an action that requires user verification, the personal information action unit (1300) can execute the action in response to user verification input. In the case of an action that requires administrator approval, the personal information action unit (1300) can generate approval request information for an administrator or report approver and execute the action according to the approval result.

[0165] If deletion is selected, the personal information action unit (1300) can delete the selected image, document, file, or email or move it to the trash. If AI automatic masking is selected, the personal information action unit (1300) can apply blur processing, mosaic processing, color replacement, character removal, area deletion, patch insertion, or other de-identification processing to the personal information area based on the location information of the personal information area generated by the on-device AI agent management unit (1500).

[0166] If encryption isolation is selected, the personal information action unit (1300) may encrypt the selected file and move it to an isolated storage area or a safe folder separated from the general storage area. If transfer to a safe folder is selected, the personal information action unit (1300) may move the selected file to a safe folder corresponding to the type of personal information or the purpose of storage. If safe destruction is selected, the personal information action unit (1300) may perform deletion processing on the selected file to restrict recovery. If exception processing is selected, the personal information action unit (1300) may exclude the image, document, file, or email from deletion, masking, isolation, or destruction, and may store at least one of the exception reason, exception validity period, exception processing user, and exception processing time as exception processing information.

[0167] In one embodiment, the personal information action unit (1300) may provide a verification request for an image or document image after masking processing to the on-device AI agent management unit (1500). The on-device AI agent management unit (1500) may reapply at least one of an optical character recognition model, an object detection model, an image classification model, or a document layout analysis model to the image after masking processing, and determine whether a personal information string, a face area, an ID card area, or document identification information is re-detected in the masked area. If the possibility of re-detection exceeds a preset standard, the on-device AI agent management unit (1500) may provide additional masking candidates to the personal information action unit (1300).

[0168] Next, The personal information action unit (1300) can provide personal information action result information to the personal information deletion and PC security setting management unit (1400) and the inspection result judgment unit (1600) (S4700).

[0169] The personal information action result information may include at least one of the following: identification information of the action target, type of action, time of action request, time of action completion, whether deletion, whether masking, whether encryption isolation, whether transfer to a safe folder, whether safe destruction, whether exception processing, reason for exception, validity period of exception, user approval, administrator approval, whether action success, and reason for action failure.

[0170] The personal information deletion and PC security setting management unit (1400) can update the personal information deletion status or storage status based on the personal information action result information. For example, if file deletion is performed, the status of the Recycle Bin can be checked, and if transfer to a safe folder is performed, the storage period, scheduled time of destruction, and remaining storage period of the file can be set, and if exception processing is performed, the exception validity period and whether the exception has expired can be managed.

[0171] Next, The inspection result judgment unit (1600) can determine whether the measure is completed based on personal information measure result information, risk score before the measure and risk score after the measure, and can generate first inspection result information corresponding to the personal information search and measure process (S4800).

[0172] The inspection result judgment unit (1600) can obtain a first risk score before the execution of the action scenario and a second risk score after the execution of the action scenario from the on-device AI agent management unit (1500). The inspection result judgment unit (1600) can calculate the amount of risk reduction by comparing the first risk score and the second risk score.

[0173] The inspection result judgment unit (1600) may determine the measure to be completed if the amount of risk reduction is greater than or equal to the standard for each type of measure, and may determine the measure to be partially taken, require further verification, or fail if the amount of risk reduction is less than the standard for each type of measure.

[0174] For example, the deletion of files containing personal information may be deemed completed if the risk score decreases below the deletion completion threshold after the action. The encryption isolation measure may be deemed completed if the storage path risk score and the external sharing possibility score decrease below the threshold. Exception handling may be deemed conditionally completed even if the risk score does not decrease, provided that the exception validity period and approver information are satisfied.

[0175] The first inspection result information may include at least one of search execution information, search result information, personal information detection result, type of personal information, sensitivity of personal information, purpose of storage, allowable storage period, whether the storage period has been exceeded, action execution information, authority delegation level, risk score before action, risk score after action, amount of risk reduction, and result of determining completion of action.

[0176] The inspection result determination unit (1600) may store the first inspection result information in conjunction with the inspection session identification information or provide it to the report generation unit (1700). At this time, the original image, document body, email body, filename, storage path, or actual personal information value containing personal information may be maintained within the user terminal and managed separately from external verification information.

[0177] Accordingly, the method according to FIG. 4 may represent a flow of personal information search and action performed by the user authentication unit (1100), personal information search unit (1200), on-device AI agent management unit (1500), personal information action unit (1300), personal information deletion and PC security setting management unit (1400), and inspection result determination unit (1600) of FIG. 3 being interconnected.

[0178] In addition, the method according to FIG. 4 is not limited to simply providing personal information search results, but can also analyze the search result information in the on-device AI agent management unit (1500) to generate the purpose of storage, whether the storage period has been exceeded, the risk score, the action scenario, and the authority delegation level, and the personal information action unit (1300) can perform actions based on the action scenario and the authority delegation level, and the inspection result judgment unit (1600) can determine whether the action is completed based on the amount of risk reduction before and after the action.

[0179] FIG. 5 is a flowchart illustrating a method in which a security inspection device according to one embodiment of the present specification checks the personal information deletion status and PC security setting status using the personal information search and action results according to FIG. 4, and generates second inspection result information based on the on-device AI agent analysis results.

[0180] Referring to Fig. 5, First, the personal information deletion and PC security setting management unit (1400) of the security inspection device (1000) can obtain personal information action result information or first inspection result information generated in FIG. 4 (S5100).

[0181] The above personal information action result information may include at least one of the following: identification information of the action target, type of action, time of action request, time of action completion, whether deletion, whether masking, whether encryption isolation, whether transfer to a safe folder, whether safe destruction, whether exception processing, reason for exception, validity period of exception, user approval, administrator approval, success of action, and reason for failure of action.

[0182] The first inspection result information above is information generated during the personal information search and action process according to FIG. 4, and may include at least one of search execution information, search result information, personal information detection result, personal information type, personal information sensitivity, storage purpose, allowable storage period, whether the storage period has been exceeded, action execution information, authority delegation level, risk score before action, risk score after action, risk reduction amount, and action completion judgment result.

[0183] The personal information deletion and PC security setting management unit (1400) may determine items for subsequent inspection based on the personal information action result information or the first inspection result information. For example, if a file deletion action is performed in FIG. 4, the personal information deletion and PC security setting management unit (1400) may determine the trash bin cleanup status or the deletion completion status as items for subsequent inspection. If a safe folder transfer action is performed in FIG. 4, the personal information deletion and PC security setting management unit (1400) may determine the retention period, scheduled time of destruction, and remaining retention period of the safe folder as items for subsequent inspection. If an exception processing is performed in FIG. 4, the personal information deletion and PC security setting management unit (1400) may determine the exception validity period and whether the exception has expired as items for subsequent inspection.

[0184] Next, The personal information deletion and PC security setting management unit (1400) can set the personal information deletion status or PC security setting status of the user terminal as a target for inspection (S5200).

[0185] The personal information deletion status may include at least one of the following: a safe folder management status, a download folder cleanup status, a trash bin cleanup status, a browser usage history deletion status, an encrypted quarantine storage area status, a safe destruction completion status, and an exception handling validity period status.

[0186] The PC security setting status may include at least one of screen saver settings, network sharing settings, operating system password settings, firewall settings, account permissions, whether unauthorized programs are installed, operating system update status, security score of an external security program, and inspection results of a personal information management program.

[0187] In one embodiment, the personal information deletion and PC security setting management unit (1400) can determine the priority of inspection target items based on the type of personal information, purpose of storage, allowed storage period, type of action, or risk score included in the first inspection result information generated in FIG. 4. For example, if a file containing an identification card or resident registration number is transferred to a safe folder, the storage period and scheduled time of destruction of the said safe folder can be set as priority inspection targets.

[0188] Next, The personal information deletion and PC security setting management unit (1400) can obtain pre-action status information for the set inspection target items (S5300).

[0189] When the safe folder management status is set to be a target for inspection, the personal information deletion and PC security setting management unit (1400) can obtain at least one of whether the safe folder is created, a list of files stored in the safe folder, the type of personal information per file, the purpose of storage per file, the storage period, the scheduled time of destruction, the remaining storage period, and whether it is a target for destruction.

[0190] When the download folder cleanup status is set as a target for inspection, the personal information deletion and PC security setting management unit (1400) can obtain at least one of the existence of a file stored in the download folder, the number of files, file name, file size, creation time, modification time, and possibility of containing personal information.

[0191] When the trash bin cleanup status is set as a target for inspection, the personal information deletion and PC security setting management unit (1400) can check at least one of the existence of files stored in the trash bin, the number of files, whether they can be recovered, and the number of deleted files that may contain personal information.

[0192] When the browser usage history deletion status is set as a target for inspection, the personal information deletion and PC security setting management unit (1400) can check the type of browser installed on the user terminal, whether the browser usage history deletion function can be called, and whether at least one of the visit history, download history, cookies, cache, and stored input information exists.

[0193] When the PC security setting status is set as a target for inspection, the personal information deletion and PC security setting management unit (1400) can obtain at least one of the following: whether a screen saver is set, the screen saver waiting time, whether a logon screen is displayed when the screen saver is disabled, whether network sharing is enabled, whether an operating system password is set, whether a firewall is enabled, whether an administrator account exists, whether an unauthorized program is installed, and the operating system update status.

[0194] Next, The personal information deletion and PC security setting management unit (1400) provides the acquired pre-action state information to the on-device AI agent management unit (1500), and the on-device AI agent management unit (1500) can generate a risk score for each inspection item and a security action scenario based on the pre-action state information and the first inspection result information (S5400).

[0195] The on-device AI agent management unit (1500) can analyze the status information prior to action to determine the vulnerability, risk score, and whether action is required for each item subject to inspection. For example, if there is a file in the download folder that is highly likely to contain personal information, if the retention period of a file in the safe folder has been exceeded, or if there is a recoverable personal information file in the trash, the on-device AI agent management unit (1500) can assign a relatively high risk score to the item for deleting personal information.

[0196] In addition, if the screen saver waiting time exceeds the standard time, the network sharing function is enabled, an unauthorized program is installed, or the operating system security update is not performed, the on-device AI agent management unit (1500) can identify the PC security setting item as a vulnerable item and calculate a risk score for the item.

[0197] The security measure scenario may include at least one of setting the safe folder retention period, setting the scheduled destruction time, deleting files from the safe folder, cleaning the download folder, emptying the trash, calling the browser history deletion screen, calling the screen saver settings screen, disabling network sharing, calling the operating system password settings screen, changing firewall settings, adjusting account permissions, calling the unauthorized program removal screen, calling the operating system update screen, running an external security program, requesting user verification, or requesting administrator approval.

[0198] In one embodiment, the on-device AI agent management unit (1500) can correct a security action scenario using the personal information sensitivity, storage purpose, allowed storage period, risk score, or action completion judgment result included in the first inspection result information generated in FIG. 4. For example, if a high-risk personal information file is determined to be an exception in FIG. 4, the on-device AI agent management unit (1500) can generate a security action scenario in FIG. 5 that includes shortening the exception validity period, requesting administrator approval, or transferring to a safe folder.

[0199] Next, The on-device AI agent management unit (1500) determines the authority delegation level of the security action scenario based on user authority information, the risk level of the inspection item, the recoverability of the action method, the organization policy, and whether administrator approval is required, and the personal information deletion and PC security setting management unit (1400) can perform personal information deletion or PC security setting measures based on the security action scenario and the authority delegation level (S5500).

[0200] The authority delegation level can be determined as any one of the following: immediate execution, user verification required, explicit approval required, pre-delegation possible, or administrator approval required. For example, calling a settings screen or providing instructions on how to take action may be classified as immediate execution; cleaning the download folder or deleting browser history may be classified as requiring user verification; and disabling network shares or changing firewall settings may be classified as pre-delegation possible or requiring explicit approval. Additionally, the safe destruction of high-risk personal information files within a safe folder or changing an account with administrator privileges may be classified as requiring administrator approval.

[0201] The personal information deletion and PC security setting management unit (1400) can perform at least one of calling a setting screen, providing user action guidance, one-click action, changing automatic settings within the scope of delegated authority, or requesting administrator approval depending on the authority delegation level.

[0202] When an action is taken regarding the personal information deletion item, the personal information deletion and PC security setting management unit (1400) can perform at least one of deleting safe folder files, updating the safe folder retention period, setting the scheduled time of destruction, organizing the download folder, emptying the trash bin, deleting browser usage history, or updating the exception validity period.

[0203] When an action is taken regarding a PC security setting item, the personal information deletion and PC security setting management unit (1400) can perform at least one of screen saver setting, network sharing setting change, operating system password setting, firewall setting change, account permission adjustment, removal of unauthorized programs, or execution of an operating system update.

[0204] Next, The personal information deletion and PC security setting management unit (1400) can re-obtain status information after action on the inspection target item after personal information deletion or PC security setting action is performed (S5600).

[0205] For example, if a download folder cleanup measure is performed, the personal information deletion and PC security setting management unit (1400) can reconfirm the number of files remaining in the download folder after the measure or the number of files that may contain personal information. If a trash bin emptying measure is performed, the existence of recoverable files stored in the trash bin can be reconfirmed. If the safe folder retention period is renewed, the renewed retention period, the scheduled time of destruction, and the remaining retention period can be reconfirmed.

[0206] If the screen saver settings are changed, the personal information deletion and PC security settings management unit (1400) can reconfirm whether the screen saver is set, the standby time, and whether the logon screen is displayed. If the network sharing settings are changed, it can reconfirm whether network search, file and printer sharing, or public folder sharing is enabled. If unauthorized program removal is performed, it can reconfirm whether the program has been removed from the list of installed programs. If an operating system update is performed, it can reconfirm whether the update has been applied or the security patch status.

[0207] The personal information deletion and PC security setting management unit (1400) can provide at least one of the state information before the action, state information after the action, security action scenario, authority delegation level, and action execution result to the inspection result judgment unit (1600).

[0208] Next, The inspection result judgment unit (1600) can compare the state information before the measure and the state information after the measure, and determine whether the measure for the personal information deletion state and PC security setting state is completed based on the risk reduction amount of the risk score before the measure and the risk score after the measure (S5700).

[0209] The inspection result judgment unit (1600) compares the standard state and the state information after action for each inspection item, and if the state information after action satisfies the standard state, the item can be judged as completed or excellent. If the state information after action satisfies only some of the standards, the item can be judged as partially taken, and if the state information after action does not satisfy the standard state, the item can be judged as requiring additional verification or as failed to take action.

[0210] Additionally, the inspection result judgment unit (1600) can calculate the amount of risk reduction by comparing the risk score before the measure and the risk score after the measure. The inspection result judgment unit (1600) can determine that the measure has been effectively performed if the amount of risk reduction is greater than or equal to the standard for each inspection item, and can determine that the measure is a partial measure or requires additional verification if the amount of risk reduction is less than the standard.

[0211] For example, if the network sharing setting is changed from an enabled state to a disabled state, and it is confirmed that the sharing function for the re-measurement result is disabled after the action, the inspection result determination unit (1600) may determine that the action is in a completed state. On the other hand, if a disabled action is requested but the sharing function for the re-measurement result is still enabled after the action, the inspection result determination unit (1600) may determine that the action is in a failed state or requires additional verification.

[0212] Next, The inspection result judgment unit (1600) can generate second inspection result information corresponding to the personal information deletion status and PC security setting status, and provide the second inspection result information to the report generation unit (1700) (S5800).

[0213] The second inspection result information may include at least one of inspection session identification information, inspection item identification information, inspection item type, status information before action, status information after action, security action scenario, authority delegation level, user approval status, administrator approval status, risk score before action, risk score after action, risk reduction amount, completion status of action, reason for judgment, and reason for action failure.

[0214] The second inspection result information may be provided to the report generation unit (1700) together with the first inspection result information generated during the personal information search and action process according to FIG. 4. The report generation unit (1700) may generate an inspection report including item-specific status, overall score, overall grade, priority action items, and security insights based on the first inspection result information and the second inspection result information.

[0215] Accordingly, the method according to FIG. 5 can input personal information action result information or first inspection result information generated in FIG. 4 to subsequently inspect the personal information deletion status and PC security setting status, and the on-device AI agent management unit (1500) generates a risk score, security action scenario, and authority delegation level for each inspection item, the personal information deletion and PC security setting management unit (1400) performs actions within the scope of user approval or delegated authority, and the inspection result judgment unit (1600) can generate second inspection result information based on the status information before and after the action and the amount of risk reduction.

[0216] In addition, the method according to Fig. 5 can subsequently verify residual risks and terminal security vulnerabilities after personal information measures by checking the personal information deletion status, retention period management, exception processing validity period, trash bin cleanup status, and PC security setting status in conjunction with the personal information search and measures results of Fig. 4 without being disconnected.

[0217] FIG. 6 is a flowchart illustrating a method for integrating first inspection result information and second inspection result information according to one embodiment of the present specification and verifying inspection results based on an action consistency score.

[0218] Referring to Fig. 6, First, the inspection result determination unit (1600) of the security inspection device (1000) can obtain first inspection result information generated during the personal information search and action process according to FIG. 4, and second inspection result information generated during the inspection process of the personal information deletion status and PC security setting status according to FIG. 5 (S6100).

[0219] The first inspection result information may include at least one of search execution information, search result information, personal information detection result, type of personal information, sensitivity of personal information, purpose of storage, allowable storage period, whether the storage period has been exceeded, action execution information, authority delegation level, risk score before action, risk score after action, amount of risk reduction, and result of determining completion of action.

[0220] The second inspection result information may include at least one of inspection session identification information, inspection item identification information, inspection item type, status information before action, status information after action, security action scenario, authority delegation level, user approval status, administrator approval status, risk score before action, risk score after action, risk reduction amount, completion status of action, reason for judgment, and reason for action failure.

[0221] The inspection result determination unit (1600) can link the first inspection result information and the second inspection result information based on at least one of the same inspection session identification information, user identification information, terminal identification information, action target identification information, or inspection item identification information.

[0222] Next, The inspection result judgment unit (1600) can generate an integrated inspection record based on the first inspection result information and the second inspection result information (S6200).

[0223] The integrated inspection record may be information that integrates personal information search items, personal information action items, personal information deletion status items, and PC security setting items into a single inspection session unit.

[0224] The integrated inspection record may include at least one of inspection session identification information, user identification information or non-identification information, terminal identification information, inspection policy version, whether personal information search was performed, whether personal information action was performed, whether personal information deletion status was checked, whether PC security setting status was checked, item-specific status information, item-specific risk score, item-specific authority delegation level, and item-specific action completion status.

[0225] In one embodiment, the integrated inspection record may store the source or detailed search information of personal information and the result information required for external verification separately. The source or detailed search information of personal information may include an image source, document body, email body, filename, storage path, or actual personal information value. The result information required for external verification may include at least one of the inspection status by item, whether the action has been completed, the amount of risk reduction, the authority delegation level, the time of inspection completion, the policy version, and integrity verification information.

[0226] Next, The inspection result judgment unit (1600) can compare the pre-action information and post-action information for each item included in the integrated inspection record (S6300).

[0227] The inspection result judgment unit (1600) can calculate the degree of risk reduction by comparing the risk score before the measure and the risk score after the measure. The degree of risk reduction can be calculated based on the difference between the risk score before the measure and the risk score after the measure, or the ratio of the reduction of the risk score after the measure to the risk score before the measure.

[0228] Additionally, the inspection result judgment unit (1600) can calculate the degree of satisfaction of the state change by comparing the state information before the action and the state information after the action. The degree of satisfaction of the state change may be a value indicating whether the state information after the action has reached the target state required by the inspection policy.

[0229] For example, in FIG. 4, encryption isolation measures for personal information files are performed, and in FIG. 5, it is confirmed that the file exists in a safe folder or isolated storage area, and the possibility of external sharing is reduced to below a standard, the inspection result determination unit (1600) can calculate a high state change satisfaction level.

[0230] In addition, if the network sharing setting is changed from an active state to a disabled state, or if the personal information file that was in the trash is deleted and the possibility of recovery is eliminated, the inspection result judgment unit (1600) may determine that the degree of satisfaction of the state change of the item meets the criteria.

[0231] Next, The inspection result judgment unit (1600) can calculate the action consistency score based on the first inspection result information and the second inspection result information (S6400).

[0232] The action consistency score may be a value indicating whether the action performed by the user or the security inspection device (1000) led to actual risk reduction and state change, whether the action was performed in accordance with the authority delegation level, and whether the integrity of the first inspection result information, the second inspection result information, or the integrated inspection record was maintained.

[0233] In one embodiment, the inspection result judgment unit (1600) can calculate the action consistency score according to the following mathematical formula 1.

[0234]

[0235] Here, V is the action consistency score, ΔR is the degree of risk reduction calculated based on the difference between the risk score before the action and the risk score after the action, S is the state change satisfaction calculated based on the comparison result between the state information before the action and the state information after the action, A is the authority suitability indicating whether the action scenario was performed according to the authority delegation level, and I may be an integrity value representing the integrity verification result of the first inspection result information, the second inspection result information, or the integrated inspection record. Additionally, a, b, c, and d may be weights for each element.

[0236] The degree of risk reduction ΔR can be calculated based on the difference between the risk score before the action and the risk score after the action, or the ratio of reduction relative to the risk score before the action. For example, if the risk score before the action is high and the risk score after the action is reduced to below the standard, the degree of risk reduction ΔR may have a high value.

[0237] The state change satisfaction S indicates whether the state information before and after the action has reached the target state required by the inspection policy. For example, if a file containing personal information is moved from a general storage area to an encrypted isolated storage area, if a file exceeding the retention period is deleted, or if network sharing settings are disabled, the state change satisfaction S may have a high value.

[0238] Authority suitability A can indicate which authority delegation level—immediately executable, user verification required, explicit approval required, pre-delegable, or administrator approval required—the action scenario corresponds to, and whether the actual action was performed according to that authority delegation level. For example, if an action requiring administrator approval is performed without administrator approval, Authority suitability A may have a low value. Conversely, if a action requiring user verification is performed after user approval input, or an action requiring administrator approval is performed after approver approval, Authority suitability A may have a high value.

[0239] The integrity value I can indicate whether the first inspection result information, the second inspection result information, or the integrated inspection record has not been altered after creation. For example, if the verification of the hash value, message authentication code, digital signature, verification token, or root hash value of the hash tree generated in correspondence with the first inspection result information, the second inspection result information, or the integrated inspection record is successful, the integrity value I can have a high value.

[0240] In one embodiment, the report generation unit (1700) or the report verification management unit (1800) may generate an inspection report, external verification information, or report delivery information based on the action consistency score and integrated inspection record calculated in S6400. In this case, integrity verification information for the report delivery information may be generated or verified separately after the report delivery information is generated.

[0241] Next, The inspection result judgment unit (1600) can determine the inspection status of each item based on the action consistency score (S6500).

[0242] The inspection result judgment unit (1600) can determine that the inspection item is in a completed state if the action consistency score V is greater than or equal to a preset completion standard. If the action consistency score V is less than the completion standard and greater than or equal to the additional verification standard, the inspection result judgment unit (1600) can determine that the inspection item is in a state requiring partial action or additional verification. If the action consistency score V is less than the additional verification standard, the inspection result judgment unit (1600) can determine that the inspection item is in a failed state.

[0243] In one embodiment, the inspection result determination unit (1600) can determine the inspection status for each item by considering not only the action consistency score but also whether the essential inspection items are satisfied. For example, if a file subject to safe destruction of personal information is not deleted, if an action requiring administrator approval is performed without approval, or if integrity verification fails, the item may be determined to require additional verification or fail to take action, even if the action consistency score is above a certain level.

[0244] The inspection result judgment unit (1600) can determine the inspection status for each item as insufficient, caution, excellent, completed, conditionally completed, partial action, additional verification required, or action failure. Additionally, the inspection result judgment unit (1600) can generate at least one of the following: the reason for judgment for each item, the action consistency score, the degree of risk reduction, the degree of satisfaction with the state change, the degree of authorization suitability, and the integrity verification result.

[0245] Next, The report generation unit (1700) can generate an inspection report based on the integrated inspection record and the inspection status by item (S6600).

[0246] The inspection report may include at least one of the following: inspection status by item, consistency score of action by item, risk score by item, risk reduction amount before and after action, satisfaction of mandatory inspection items, overall score, overall grade, priority action items, security insights, and inspection completion time.

[0247] The report generation unit (1700) can calculate a comprehensive score or a comprehensive grade based on the inspection status and action consistency score for each item. For example, if all mandatory inspection items are in a completed state and the action consistency score of high-risk items is above the standard, the report generation unit (1700) can assign a high comprehensive grade. On the other hand, if high-risk personal information items are in an unhandled state or if mandatory items among PC security setting items do not meet the standard, the report generation unit (1700) can assign a low comprehensive grade.

[0248] Additionally, the report generation unit (1700) can determine priority action items based on risk level, necessity, degree of exceeding the retention period, possibility of external leakage, or action consistency score among items judged to be insufficient, caution, partial action, or require additional verification.

[0249] Next, The report verification management department (1800) can generate external verification information based on the inspection report (S6700).

[0250] External verification information may include at least one of user identification information or non-identification information, terminal identification information, inspection status by item, whether action is completed, action consistency score, degree of risk reduction, overall score, overall grade, inspection completion time, policy version, authority delegation level, approval or rejection status, and integrity verification information.

[0251] The report verification management unit (1800) may exclude original personal information or detailed search information from external verification information. For example, filenames, storage paths, email subject lines, email body text, original images, or actual personal information values ​​may be maintained within the user terminal, and external verification information may only include result information such as item-specific status, whether action has been completed, degree of risk reduction, authority delegation level, and integrity verification information.

[0252] The report verification management unit (1800) can generate external verification information in the form of a copyable string, an optically readable code such as a QR code, a verification token, or link information. Additionally, the report verification management unit (1800) can include creation time, expiration time, and integrity verification information in the external verification information.

[0253] The integrity verification information may include at least one of a hash value, a message authentication code, an electronic signature, encryption information, a verification token, or the root hash value of a hash tree. The report verification management unit (1800) verifies the validity period and integrity of the external verification information and may allow the external verifier to display the inspection result or provide approval input only if the verification is successful.

[0254] Next, The report verification management unit (1800) can generate approval information based on the approval or rejection input of an external verifier (S6800).

[0255] After verifying the external verification information, the external verifier may approve or reject the inspection results. The report verification management unit (1800) may generate approval information including at least one of approver information, approval time, approval status, rejection status, and reason for rejection based on the external verifier's approval or rejection input.

[0256] In one embodiment, the report verification management unit (1800) may control whether an external verifier can approve the report based on the action consistency score, the integrity verification result, or whether the required inspection items are satisfied. For example, approval input may be restricted for inspection reports that failed integrity verification, and if high-risk items remain in a failed action state, input of the reason for rejection may be required.

[0257] Accordingly, the method according to FIG. 6 can integrate the first inspection result information generated in FIG. 4 and the second inspection result information generated in FIG. 5, and verify the inspection results based on an action consistency score that reflects the degree of risk reduction, the degree of satisfaction of state change, the degree of authority suitability, and the integrity value.

[0258] In addition, the method according to Fig. 6 allows an external verifier to verify whether personal information measures and PC security setting measures have been completed, whether measures have been taken within the scope of delegated authority, whether risks have been reduced before and after the measures, and the integrity of the inspection report, without providing the original personal information or detailed search information to the outside.

[0259] Examples :

[0260] Meanwhile, FIGS. 4 to 6 are not limited to separate procedures performed independently of each other, but can be understood as linked procedures performed sequentially within a single inspection session, or at least some steps performed in parallel.

[0261] Specifically, the personal information search and action procedure according to FIG. 4 may be a procedure in which a user authentication unit (1100), a personal information search unit (1200), an on-device AI agent management unit (1500), a personal information action unit (1300), a personal information deletion and PC security setting management unit (1400), and an inspection result determination unit (1600) are interconnected and performed. The personal information search unit (1200) identifies at least one of images, documents, PC files, downloaded files, email bodies, email attachments, or browser stored data that is stored on or accessible through the user terminal as the data to be searched, and the on-device AI agent management unit (1500) may perform personal information detection, context analysis, inference of storage purpose, determination of whether the storage period has been exceeded, and calculation of a risk score based on the search result information.

[0262] Additionally, the on-device AI agent management unit (1500) may generate an action scenario including at least one of deletion, AI automatic masking, manual masking, encryption isolation, safe folder transfer, safe destruction, exception handling, user verification request, or administrator approval request based on the personal information search results and risk score of FIG. 4. The personal information action unit (1300) performs personal information action based on the action scenario and authority delegation level, and the inspection result determination unit (1600) may generate first inspection result information including the amount of risk reduction and whether the action is completed by comparing the risk score before the action and the risk score after the action.

[0263] The procedure for checking the personal information deletion status and PC security setting status according to FIG. 5 can be performed by inputting the personal information action result information or the first inspection result information generated in FIG. 4. The personal information deletion and PC security setting management unit (1400) can set at least one of the following as a subsequent inspection target based on the action result of FIG. 4: the safe folder management status, the download folder cleanup status, the trash bin cleanup status, the browser usage history deletion status, the encrypted isolated storage area status, the safe destruction completion status, or the exception processing validity period status. Additionally, the personal information deletion and PC security setting management unit (1400) can set at least one of the following as an inspection target for the PC security setting status: screen saver settings, network sharing settings, operating system password settings, firewall settings, account permissions, whether unauthorized programs are installed, operating system update status, the security score of an external security program, or the inspection result of a personal information management program.

[0264] The on-device AI agent management unit (1500) can generate risk scores for each inspection item, security action scenarios, and authority delegation levels based on the state information before action obtained in FIG. 5 and the first inspection result information in FIG. 4. The personal information deletion and PC security setting management unit (1400) can perform at least one of user action guidance, calling a settings screen, one-click action, automatic setting change within the scope of delegated authority, or requesting administrator approval based on the security action scenario and authority delegation level. The inspection result determination unit (1600) can compare the state information before action and the state information after action, and generate second inspection result information corresponding to the personal information deletion state and PC security setting state based on the risk reduction amount of the risk score before action and the risk score after action.

[0265] The inspection result verification procedure according to FIG. 6 may be a procedure for verifying whether the results of the personal information search and action, the personal information deletion status, and the PC security setting status are consistent with each other by integrating the first inspection result information generated in FIG. 4 and the second inspection result information generated in FIG. 5. The inspection result determination unit (1600) may link the first inspection result information and the second inspection result information based on at least one of the same inspection session identification information, user identification information, terminal identification information, action target identification information, or inspection item identification information, and generate an integrated inspection record that integrates the personal information search item, personal information action item, personal information deletion status item, and PC security setting item into a single inspection session unit.

[0266] The inspection result determination unit (1600) can calculate the degree of risk reduction and the degree of satisfaction for state change by comparing the information before and after the action for each item included in the integrated inspection record. Additionally, the inspection result determination unit (1600) can calculate an action consistency score by considering the degree of authority suitability according to the authority delegation level and the integrity value of the first inspection result information, the second inspection result information, or the integrated inspection record together. Accordingly, the inspection result determination unit (1600) can comprehensively determine whether the action taken by the user or the security inspection device (1000) led to actual risk reduction and state change, whether the action was performed in accordance with the authority delegation level, and whether the integrity of the inspection result information was maintained.

[0267] In one embodiment, when a personal information file is recorded as deleted in FIG. 4, FIG. 5 can subsequently check whether the file remains in the Recycle Bin, Downloads folder, or original storage area, and FIG. 6 can determine whether the deletion action is completed by using the deletion record, the actual deletion status, the degree of risk reduction, and the action consistency score together.

[0268] In one embodiment, when a high-risk personal information file in FIG. 4 is transferred to a safe folder or encrypted and isolated, FIG. 5 can check the retention period, scheduled time of destruction, remaining retention period, safe folder storage status, or isolated storage status of the file. FIG. 6 can determine whether the measure was effectively performed by utilizing the change in the file's storage location, whether the possibility of external sharing is reduced, the degree of risk reduction before and after the measure, and the suitability of the measure based on the level of authority delegation.

[0269] In one embodiment, when a personal information item is exception-processed in FIG. 4, FIG. 5 may subsequently check the exception validity period, whether the exception has expired, and the exception processing status. FIG. 6 may determine whether the item is conditionally completed, requires further verification, or failed to take action by using whether the exception processing was performed within the scope of user authority or administrator approval, whether the exception validity period meets the criteria, whether the residual risk is at an acceptable level, and whether the integrity of the exception processing information is maintained.

[0270] In one embodiment, the procedure according to FIGS. 4 to 6 may be performed to restrict the external transmission of the original personal information or detailed search information. The original personal information, document body, email body, image original, filename, storage path, or actual personal information value may be maintained within the user terminal, and the report generation unit (1700) or the report verification management unit (1800) may generate an inspection report, external verification information, or report delivery information using only the result information required for external verification. The result information required for external verification may include at least one of the item-specific inspection status, whether the action has been completed, the degree of risk reduction, the action consistency score, the authority delegation level, the overall score, the overall grade, the policy version, and the integrity verification information.

[0271] In one embodiment, the report verification management unit (1800) generates external verification information or report delivery information based on the action consistency score and inspection report of FIG. 6, and can enable an external verifier to verify whether personal information measures and PC security setting measures have been completed, whether measures have been taken within the scope of delegated authority, whether risks have been reduced before and after the measures, and the integrity of the inspection results, even without verifying the original personal information.

[0272] In one embodiment, each of the procedures of FIGS. 4 to 6 may be performed in an on-device environment of a user terminal. In this case, the on-device AI agent management unit (1500) may perform at least some of the processing required for personal information detection, context-based storage purpose inference, determination of whether the storage period has been exceeded, risk score calculation, creation of action scenarios, determination of authority delegation level, and determination of action consistency within the user terminal. Accordingly, the security inspection device (1000) can manage the personal information protection measures and PC security setting measures in a verifiable form without transmitting the original personal information to a central server.

[0273] Additionally, the procedure according to FIGS. 4 to 6 may be performed repeatedly. For example, the security inspection device (1000) may perform the procedure according to FIGS. 4 to 6 again in response to at least one of a regular inspection cycle, a user request, a change in the organization's security policy, detection of high-risk personal information, expiration of the retention period, detection of an external transmission / reception event, or an administrator request. When performed repeatedly, the first inspection result information, the second inspection result information, the integrated inspection record, the action consistency score, or the approval information generated in the previous inspection session may be used to determine the risk level, generate action scenarios, determine exception handling, or determine priority inspection items for the subsequent inspection session.

[0274] Accordingly, the security inspection device (1000) according to the present embodiment can process personal information search, personal information measures, personal information deletion status check, PC security setting check, comparison of status before and after measures, verification of authorization delegation suitability, generation of inspection report, and generation of external verification information as linked data flows within a single inspection session. In addition, the security inspection device (1000) can generate security inspection results based on actual terminal status changes, degree of risk reduction, authorization delegation level, and integrity verification results, rather than simple user confirmation input or formal check results.

[0275] The foregoing specification may be implemented as computer-readable code on a medium on which a program is recorded. A computer-readable medium includes all types of recording devices in which data that can be read by a computer system is stored. Examples of computer-readable media include Hard Disk Drives (HDDs), Solid State Disks (SSDs), ROMs, RAMs, CD-ROMs, magnetic tapes, floppy disks, optical data storage devices, etc., and also include implementations in the form of carrier waves (e.g., transmission over the Internet). Accordingly, the above detailed description should not be interpreted restrictively in all respects and should be considered exemplary. The scope of this specification should be determined by a reasonable interpretation of the appended claims, and all modifications within the equivalent scope of this specification are included within the scope of this specification.

[0276] Furthermore, although the above description has focused on the services and embodiments, this is merely illustrative and does not limit the scope of this specification. Those skilled in the art will understand that various modifications and applications not exemplified above are possible without departing from the essential characteristics of the services and embodiments. For example, each component specifically shown in the embodiments may be modified and implemented. Differences related to such modifications and applications should be interpreted as being included within the scope of this specification as defined in the appended claims.

Claims

Claim 1 A method for personal information protection and security inspection based on an on-device AI agent performed by a security inspection device, comprising: a user authentication unit that authenticates a user based on user authentication information and generates user authorization information for the authenticated user; a personal information search unit that sets a personal information search type including personal information image search, PC file search, and internal email search based on the user authorization information and sets search conditions corresponding to the personal information search type; and a step in which the personal information search unit identifies search target data stored in a user terminal or accessible through the user terminal based on the personal information search type and the search conditions, and generates search result information for the search target data. A method comprising: a search condition including a search target drive, a search target path, whether subfolders are included, a file extension, a filename, a search keyword, a creation time, a modification time, a last access time, a search target mailbox, a search period, a sender, a recipient, whether the email body is searched, and whether attachments are searched; the search target data including images, documents, PC files, downloaded files, compressed files, email bodies, email attachments, and browser saved data; and the search result information including a filename, a save path, a file format, a file size, a creation time, a modification time, an email subject, an attachment filename, an email receipt date, and whether user-specified keywords are included. Claim 2 The method according to claim 1 further comprises: an on-device AI agent management unit performing personal information detection, context analysis, inference of storage purpose, determination of whether the storage period has been exceeded, and calculation of a risk score based on the search result information, and generating a personal information action scenario and an authority delegation level based on the risk score; a personal information action unit performing personal information action based on the personal information action scenario and the authority delegation level and generating personal information action result information; and an inspection result judgment unit generating a first inspection result information corresponding to the personal information search and action process based on the personal information action result information, the risk score before action, and the risk score after action. Claim 3 In paragraph 2, the method further comprises the step of a personal information deletion and PC security setting management unit acquiring the personal information action result information or the first inspection result information, and determining subsequent inspection target items for the personal information deletion status or PC security setting status based on the personal information action result information or the first inspection result information; wherein the subsequent inspection target items include a safe folder management status, a download folder cleanup status, a trash bin cleanup status, a browser usage history deletion status, an encrypted isolated storage area status, a safe destruction completion status, an exception handling validity period status, a screen saver setting, a network sharing setting, an operating system password setting, a firewall setting, account authority, whether an unauthorized program is installed, and an operating system update status. Claim 4 A method further comprising: a step in which the personal information deletion and PC security setting management unit acquires state information prior to action regarding the item subject to subsequent inspection and provides the state information prior to action to the on-device AI agent management unit; a step in which the on-device AI agent management unit generates a risk score, a security action scenario, and an authority delegation level for each inspection item based on the state information prior to action and the first inspection result information; and a step in which the personal information deletion and PC security setting management unit performs a personal information deletion measure or a PC security setting measure based on the security action scenario and the authority delegation level. Claim 5 In claim 4, the method further comprises: a step in which the personal information deletion and PC security setting management unit re-acquires post-action status information for the subsequent inspection target item after the personal information deletion measure or the PC security setting measure is performed, and provides the pre-action status information, the post-action status information, the security measure scenario, the authority delegation level, and the result of performing the measure to the inspection result judgment unit; and a step in which the inspection result judgment unit compares the pre-action status information and the post-action status information, and generates second inspection result information corresponding to the personal information deletion status and the PC security setting status based on the risk reduction amount of the risk score before the measure and the risk score after the measure. Claim 6 A method comprising: a step in which the inspection result determination unit links the first inspection result information and the second inspection result information based on the same inspection session identification information, user identification information, terminal identification information, action target identification information, or inspection item identification information, and generates an integrated inspection record by integrating personal information search items, personal information action items, personal information deletion status items, and PC security setting items into a single inspection session unit; a step in which the inspection result determination unit calculates the degree of risk reduction and the degree of satisfaction of state change by comparing information before and after action for each item included in the integrated inspection record; a step in which the inspection result determination unit calculates an action consistency score based on the degree of risk reduction, the degree of satisfaction of state change, the degree of authority suitability according to the authority delegation level, and the integrity value of the first inspection result information, the second inspection result information, or the integrated inspection record; and a step in which the inspection result determination unit determines the inspection status for each item based on the action consistency score. Claim 7 A security inspection device for performing personal information protection and security inspection based on an on-device AI agent comprises: a user authentication unit that authenticates a user based on user authentication information and generates user authorization information for the authenticated user; and a personal information search unit that sets a personal information search type including personal information image search, PC file search, and internal email search based on the user authorization information, sets search conditions corresponding to the personal information search type, identifies search target data stored in a user terminal or accessible through the user terminal based on the personal information search type and the search conditions, and generates search result information for the search target data. A security inspection device comprising, wherein the search conditions include a search target drive, a search target path, whether subfolders are included, a file extension, a filename, a search keyword, a creation time, a modification time, a last access time, a search target mailbox, a search period, a sender, a recipient, whether the email body is searched, and whether attachments are searched, wherein the search target data includes images, documents, PC files, downloaded files, compressed files, email bodies, email attachments, and browser saved data, and wherein the search result information includes a filename, a save path, a file format, a file size, a creation time, a modification time, an email subject, an attachment filename, an email receipt date, and whether user-specified keywords are included.

Citation Information

Patent Citations

  • Electronic mail device and electronic mail program

    JP2006201836A

  • Method and server for comparing commodities

    KR1020100045113A

  • Method and apparatus for managing personal information access records

    KR1020240036218A

  • File search system and file search server device

    US20100106709A1