Method and device for improving security in using chatgpt based on messenger program

Abstract

An embodiment of the present invention relates to a method and device for improving security in using chatGPT based on a messenger program. The embodiment relates to a method and device for improving security in using chatGPT based on a messenger program, wherein security regulations are applied to messages to prevent information requiring security from being leaked to a chatGPT server.

Description

Method and device for improving security using ChatGPT using a messenger program

[0001] The present invention relates to a method and device for improving security of ChatGPT using a messenger program, and more particularly, to a method and device for improving security of ChatGPT using a messenger program for preventing information requiring security from being leaked to ChatGPT when using a messenger program linked to ChatGPT.

[0002] ChatGPT is a conversational AI service based on GPT-3.5 and GPT-4. Prior to ChatGPT, conversational AI often failed to properly understand human speech or provide responses that matched the user's intent.

[0003] For conversational AI to engage in human-like conversations, it must not only possess knowledge but also understand human language. Existing conversational AI has often fallen short in this area, often responding incoherently or failing to respond at all. This has severely limited its utility beyond short-answer questions.

[0004] However, unlike existing conversational AI, ChatGPT understands the context of speech and remembers previous conversations, reaching a level where it can understand the meaning of a conversation even when inputting it as if it were a human conversation. Furthermore, ChatGPT goes beyond simply understanding the meaning of a conversation, and even answers questions about specialized knowledge at a level that surpasses the average human's ability. Consequently, interest in ChatGPT has grown worldwide, and ChatGPT is being utilized in a variety of fields.

[0005] However, as ChatGPT is utilized across diverse fields and business sectors, security concerns and corporate information leaks are emerging. As ChatGPT is used to enhance work capabilities, there are instances of sensitive information, such as confidential information or personal information, being entered into ChatGPT.

[0006] Questions entered into ChatGPT can be used for future AI training, and theoretically, Open AI operators can read all of the content. Furthermore, if Open AI is hacked, there's a risk that this information could be leaked. Therefore, if sensitive information such as personal information, confidential corporate information, or confidential documents are entered into ChatGPT, there's a risk that this sensitive information could be leaked.

[0007] A method and device for improving security of ChatGPT using a messenger program according to an embodiment of the present invention is intended to increase convenience of using ChatGPT when using a messenger program.

[0008] In addition, the method and device for improving security using ChatGPT using a messenger program according to an embodiment of the present invention is for improving security when using ChatGPT.

[0009] However, the technical task that this embodiment seeks to achieve is not limited to the technical task described above, and other technical tasks may exist.

[0010] As a technical means for achieving the above-described technical task, a method for improving security using ChatGPT using a messenger program according to an embodiment of the present invention includes a step of setting security regulations and security information of a messenger program, a step of performing security filtering of a message using a messenger program learned based on the security regulations and the security information when a user inputs a message, a step of transmitting a message on which the security filtering is completed to a ChatGPT server, and a step of receiving a response to the message from the ChatGPT server and providing the response to the message to the user.

[0011] In addition, the step of setting the security regulations and the security information according to the embodiment of the present invention includes the step of setting the security level of the security regulations and the security information, and the step of setting the application target for each security level.

[0012] In addition, the step of setting the security regulations and the security information according to an embodiment of the present invention further includes the step of setting information regarding at least one of security regulations for each security level, security keywords, security information patterns, and security documents.

[0013] In addition, the step of performing security filtering of a message according to an embodiment of the present invention includes the step of performing context and word learning on at least one of security regulations, security keywords, security information patterns, and security documents for each security level using the messenger program learned using BERT (Bidirectional Encoder Representations from Transformers), and the step of performing security filtering on the message based on the learning result.

[0014] In addition, the step of performing security filtering of a message according to an embodiment of the present invention further includes the step of replacing the first security keyword with a substitute word when the message includes a first security keyword, and the step of blocking transmission of the message so that the message is not transmitted to the ChatGPT server when the message includes a second security keyword.

[0015] In addition, the step of performing security filtering of a message according to an embodiment of the present invention further includes the step of replacing the character corresponding to the security information pattern with a substitute character when the message includes a character corresponding to the security information pattern.

[0016] In addition, the step of performing security filtering of a message according to an embodiment of the present invention further includes the step of deriving a first similarity between the message and the security document, and the step of blocking transmission of the message so that the message is not transmitted to the ChatGPT server if the first similarity is greater than or equal to a first reference similarity.

[0017] In addition, the step of performing security filtering of a message according to an embodiment of the present invention further includes the steps of creating a secure messenger channel, creating a snapshot of some or all of the messages occurring in the secure messenger channel, and performing security learning of the messenger program based on the snapshot.

[0018] In addition, the step of performing security filtering of a message according to an embodiment of the present invention further includes the step of performing word and context analysis of the message, and the step of performing security filtering of the message based on the word and context analysis result included in the message and at least one of the security regulations, security keywords, security information patterns and security documents for each security level, and if the security level of the message is determined to be a message that violates the security level set for each user, the step of blocking transmission of the message so that the message is not transmitted to the ChatGPT server.

[0019] In addition, the step of performing security filtering of a message according to an embodiment of the present invention further includes the step of generating and displaying a notification including information that security filtering has been performed on the message because the message contains information that violates security regulations when security filtering of the message is performed, and the step of generating an administrator notification including the message and information on the security filtering performed when the security filtering is performed, and providing the administrator notification to an administrator account.

[0020] In addition, a device for enhancing security using ChatGPT using a messenger program according to an embodiment of the present invention includes a communication module for transmitting and receiving information between a terminal and the messenger server and the ChatGPT server, a memory for storing at least one instruction, and a processor for executing the at least one instruction, wherein the processor sets security regulations and security information of the messenger program by executing the at least one instruction, and when a user inputs a message, performs security filtering of the message using the messenger program learned based on the security regulations and the security information, transmits the message on which the security filtering is completed to the ChatGPT server, and receives a response to the message from the ChatGPT server and provides the response to the message to the user.

[0021] A method and device for improving security of ChatGPT using a messenger program according to an embodiment of the present invention can increase convenience of using ChatGPT when using a messenger program.

[0022] In addition, the method and device for improving security using ChatGPT using a messenger program according to an embodiment of the present invention can improve security when using ChatGPT.

[0023] Figure 1 is an exemplary diagram showing a communication connection of a security enhancement device using a chatGPT messenger program according to an embodiment of the present invention.

[0024] Figure 2 is a configuration diagram of a messenger server according to an embodiment of the present invention.

[0025] Figure 3 is a configuration diagram of a terminal according to an embodiment of the present invention.

[0026] Figure 4 is a conceptual diagram illustrating the function of a processor according to an embodiment of the present invention.

[0027] Figure 5 is a flowchart of a method for improving security using ChatGPT using a messenger program according to an embodiment of the present invention.

[0028] Below, with reference to the attached drawings, embodiments of the present invention are described in detail so that those skilled in the art can easily implement them. However, the present invention may be implemented in various different forms and is not limited to the embodiments described herein. In the drawings, irrelevant parts have been omitted for clarity of description, and similar reference numerals have been used throughout the specification to indicate similar elements.

[0029] Throughout the specification, when a part is said to be "connected" to another part, this includes not only the cases where the parts are "directly connected" but also the cases where the parts are "electrically connected" with other elements intervening. Furthermore, when a part is said to "include" a component, this does not exclude other components, but rather includes other components, unless otherwise stated.

[0030] In addition, the attached drawings are only intended to facilitate easy understanding of the embodiments disclosed in this specification, and the technical ideas disclosed in this specification are not limited by the attached drawings, and should be understood to include all modifications, equivalents, or substitutes included in the spirit and technical scope of the present invention.

[0031] Terms that include ordinal numbers, such as first, second, etc., may be used to describe various components, but the components are not limited by these terms. These terms are used solely to distinguish one component from another.

[0032] When a component is referred to as being "connected" or "connected" to another component, it should be understood that it may be directly connected or connected to that other component, but that there may be other components intervening. Conversely, when a component is referred to as being "directly connected" or "connected" to another component, it should be understood that there are no other components intervening.

[0033] Singular expressions include plural expressions unless the context clearly indicates otherwise.

[0034] In this application, terms such as “include” or “have” are intended to specify the presence of a feature, number, step, operation, component, part or combination thereof described in the specification, but should be understood not to exclude in advance the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts or combinations thereof.

[0035] Hereinafter, a device for providing a private chat service according to an embodiment of the present invention will be described with reference to FIG. 1.

[0036] FIG. 1 is an exemplary diagram showing a communication connection between a messenger server (100), a terminal (200), and a ChatGPT server (300) according to an embodiment of the present invention.

[0037] Referring to FIG. 1, a messenger server (100) is connected to a terminal (200) through a communication network, and the messenger server (100) is connected to a chatGPT server (300) through the communication network. At this time, the messenger server (100) refers to a device for providing message transmission and reception or chatbot functions between users using specific services or applications, such as an in-house chatbot service, a messenger service, an in-house messenger service, a messenger service within a fan platform, etc., and may correspond to a fan platform server, a social network service server, etc.

[0038] The terminal (200) includes a terminal of a user using a messenger service or messenger program. The terminal (200) may refer to a device that displays messages and notifications received from the messenger server (100) to the user.

[0039] The messenger server (100) can transmit a message entered by a user using a terminal (200) to the chat GPT server (300). In addition, the chat GPT server (300) can receive a response to the user's message and provide it to the terminal (200).

[0040] At this time, the messenger server (100) filters whether the message entered by the user contains information requiring security, or filters messages that violate security regulations, thereby preventing messages and information requiring security from being transmitted to the chat GPT server (300). In addition, the messenger server (100) can generate a list of messages and notifications and provide them to the user. The specific configuration of the messenger server (100) will be described in detail with reference to FIGS. 2 and 4 described below.

[0041] The terminal (200) may refer to any type of handheld-based wireless communication device such as a notebook, desktop, laptop, wireless communication device with portability and mobility, or a smartphone, tablet PC, etc. equipped with a web browser.

[0042] In addition, the communication network illustrated in FIG. 1 can be implemented as a wired network such as a Local Area Network (LAN), a Wide Area Network (WAN), or a Value Added Network (VAN), or any type of wireless network such as a mobile radio communication network or a satellite communication network.

[0043] Hereinafter, the structure of a messenger server according to an embodiment of the present invention will be described with reference to FIG. 2.

[0044] FIG. 2 is a structural diagram illustrating the structure of a messenger server (100) according to an embodiment of the present invention.

[0045] Referring to FIG. 2, the messenger server (100) includes a communication module (110), a memory (120), and a processor (140), and may further include a database (130). The communication module (110) performs information transmission and reception with terminals (200, 201). The communication module (110) may include a device including hardware and software necessary for transmitting and receiving signals, such as control signals or data signals, using wired or wireless connections with other network devices.

[0046] The memory (120) stores a messenger program. The name of the messenger program is set for convenience of explanation, and the name itself does not limit the program's function. The memory (120) can store at least one of information and data input to the communication module (110), information and data required for functions performed by the processor (140), and data generated by the execution of the processor (140).

[0047] Memory (120) should be interpreted as a general term for a non-volatile storage device that maintains stored information even when no power is supplied, and a volatile storage device that requires power to maintain the stored information. In addition, memory (120) may perform a function of temporarily or permanently storing data processed by the processor (140). Memory (120) may include magnetic storage media or flash storage media in addition to volatile storage devices that require power to maintain stored information, but the scope of the present invention is not limited thereto.

[0048] The database (130) can store data related to the messenger program, such as security setting information, security document information, user information, usage history data, etc. The database (130) can constitute a part of the memory (120), but is not necessarily located within the messenger server (100), and can be connected to the outside of the messenger server (100) to perform data transmission and reception using a communication connection.

[0049] The processor (140) is configured to execute a private chat program in the memory (120). The processor (140) may include various types of devices that control and process data. The processor (140) may refer to a data processing device built into hardware that has a physically structured circuit to perform a function expressed by a code or command included in the program. In one example, the processor (140) may be implemented in the form of a microprocessor, a central processing unit (CPU), a processor core, a multiprocessor, an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), etc., but the scope of the present invention is not limited thereto.

[0050] The processor (140) is a processor that executes at least one instruction and is configured to execute a private chat program to perform the following functions and procedures.

[0051] The processor (140) sets the security rules and information of the messenger program, and when a user inputs a message, performs security filtering of the message using the messenger program learned based on the security rules and information. The processor (140) transmits the message for which security filtering has been completed to the ChatGPT server (300), and can receive a response to the message from the ChatGPT server (300) and provide it to the user.

[0052] Additionally, the processor (140) can set the target of application of security regulations and information, and set a security level for each target of application. The processor (140) can set information regarding one or more of security regulations, security keywords, security information patterns, and security documents for each security level.

[0053] In addition, the processor (140) can perform context and word learning on at least one of the security regulations, security keywords, security information patterns, and security documents for each security level using the messenger program learned using BERT (Bidirectional Encoder Representations from Transformers), and perform security filtering on the message based on the learning results.

[0054] Additionally, if the message includes a first security keyword, the processor (140) may replace the first security keyword with a substitute word and transmit the replaced message to the ChatGPT server (300). Additionally, if the message includes a second security keyword, the processor (140) may block transmission of the message so that the message is not transmitted to the ChatGPT server.

[0055] Additionally, if the message includes a character corresponding to a security information pattern, the processor (140) may replace the character corresponding to the security information pattern with a replacement character and transmit the replaced message to the chat GPT server (300).

[0056] Additionally, the processor (140) can derive a first similarity between a message and a security document, and if the first similarity is greater than or equal to a first reference similarity, the transmission of the message can be blocked so that the message is not transmitted to the ChatGPT server (300).

[0057]

[0058] Additionally, the processor (140) can create a secure messenger channel and create snapshots of some or all of the messages generated in the secure messenger channel. Furthermore, the processor (140) can perform security learning of the messenger program based on the snapshots.

[0059] Additionally, the processor (140) performs word and context analysis of the message and derives a second similarity between the word and context analysis results contained in the message and one or more of security regulations, security keywords, security information patterns, and security documents by security level. Furthermore, if the second similarity is greater than or equal to a second reference similarity, the processor (140) can block the transmission of the message so that the message is not transmitted to the ChatGPT server (300).

[0060] Additionally, the processor (140) can generate and display a message for reviewing the transmission of a message when security filtering of the message is performed, and can send a notification to an administrator account when security filtering is performed.

[0061] Figure 3 is a block diagram illustrating the configuration of a terminal (200).

[0062] Referring to FIG. 3, the terminal (200) includes a memory (220), an input / output module (230), and a processor (240), and may further include a communication module (210).

[0063] The communication module (210) can transmit and receive information with an external database or external device. Here, the external device may be the messenger server (100) described above. The memory (220) stores a messenger program. The name of the messenger program is set for convenience of explanation, and the name itself does not limit the program's functions.

[0064] The input / output module (230) can receive information, data, etc. transmitted from the outside to the terminal (200), or output information, data, etc. held by the terminal (200) to the outside. For example, the input / output module (230) can include a display, a touchpad, etc. The processor (240) executes a private chat program stored in the memory (220). Additional descriptions of the communication module (210), the memory (220), and the processor (240) will be replaced with descriptions of the communication module (110 of FIG. 2), the memory (120 of FIG. 2), and the processor (140 of FIG. 2) previously described with reference to FIG. 2.

[0065] Additionally, the processor (240) is configured to execute a messenger program to perform the following functions and procedures.

[0066] The processor (240) can transmit artist messages or fan messages generated in the terminal (200) to the messenger server (100). In addition, the processor (240) can generate and display messages, feeds, alarms, interfaces, etc. based on data received from the messenger server (100).

[0067] Hereinafter, the function and operation of the processor (140) according to an embodiment of the present invention will be described in detail with reference to FIG. 4.

[0068] Figure 4 is a conceptual diagram illustrating the function of a messenger server according to an embodiment of the present invention.

[0069] Referring to FIG. 4, the processor (140) may include a security setting module (141), a security message filtering module (142), and a notification generation module (143). The security setting module (141) may set security criteria for filtering content requiring security, such as security regulations and information.

[0070] Specifically, the security setting module (141) can set security regulations and security information for each security level, and can set the target of application of the security level. In addition, the security setting module (141) can set the target of application of the security regulations, and can set different security levels for each target of application of the security regulations. The security setting module (141) can set security management information regarding security filtering criteria, such as security regulations, security keywords, security information patterns, and security documents, and can set different security management information for each security level.

[0071] The security message filtering module (142) performs learning to filter whether a message entered by a user complies with security regulations and includes security information based on the security management information set in the security setting module (141).

[0072] The security message filtering module (142) can perform natural language processing using a language model such as a Large Language Model (LLM). In addition, the security message filtering module (142) continuously learns whether a user's input message complies with security regulations and contains security information by repeatedly performing pre-training and fine-tuning on a data set of security management information based on a BERT (Bidirectional Encoder Representations from Transformers) model.

[0073] In addition, the security message filtering module (142) can convert the user's input message filtering results into a learnable dataset and perform learning using a BERT (Bidirectional Encoder Representations from Transformers) model. The structure of the security message filtering module (142) is not limited to the BERT model, and can be input into a deep learning structure such as a recurrent neural network (RNN) or a convolutional neural network (CNN) to perform keyword-by-keyword classification of fan messages.

[0074] When using a convolutional neural network (CNN) structure, the convolutional neural network (CNN) structure can be formed to include a sentence matrix, a convolutional layer, a pooling layer, and a softmax function.

[0075] The security message filtering module (142) filters the user's input message based on security regulations by security level, question-and-answer monitoring evaluation, security keywords, security information patterns, and question-and-answer filtering results, thereby preventing information requiring security from being leaked to the ChatGPT server (300).

[0076] The security message filtering module (142) can perform segmentation of security regulations based on the user's input message and the response information of the chat GPT server (300) to the input message.

[0077] The security message filtering module (142) can derive a level of security compliance by evaluating whether a message entered by a user and a response from the ChatGPT server comply with security regulations.

[0078] Additionally, the security message filtering module (142) may provide a monitoring interface that displays information about users' input messages so that an administrator can monitor the users' input messages. The monitoring interface may include an area that displays the validity and security compliance of the filtering results of the security message filtering module (142).

[0079] In addition, the security message filtering module (142) can generate and provide a filtering information provision interface that provides the user with information about the filtered content and the filtering reason when filtering is performed on the user's input message.

[0080] Additionally, the security message filtering module (142) can determine whether the user's input message contains a security keyword, and if the user's input message contains a security keyword, can replace the security keyword with a substitute word or delete the security keyword. In addition, the security message filtering module (142) can perform filtering to prevent the user's input message containing the security keyword from being transmitted to the ChatGPT server (300).

[0081] Specifically, the security message filtering module (142) can replace the first security keyword with a substitute word if the user's input message includes the first security keyword. In addition, the security message filtering module (142) can block the transmission of the user's input message if the user's input message includes the second security keyword so that the user's input message including the second security keyword is not transmitted to the ChatGPT server (300).

[0082] Accordingly, in the case of keywords with low security levels, conversion to alternative keywords is performed, but in the case of keywords with high security levels, the message itself can be prevented from being transmitted to the ChatGPT server (300). In this case, the second security keyword may refer to a keyword set to correspond to a security level higher than that of the first security keyword.

[0083] In addition, if the user's input message includes characters corresponding to a security information pattern, the security message filtering module (142) can replace the characters corresponding to the security information pattern with replacement characters and transmit the user's input message to the chatGPT server (300).

[0084] At this time, the security information pattern means information having certain rules and patterns, such as a phone number, resident registration number, address, password, etc., and by filtering the security information having such patterns, information requiring security can be prevented from being transmitted to the ChatGPT server (300).

[0085] Additionally, the security message filtering module (142) derives the similarity between the user's input message or the user's attached data and the security document. If the derived similarity is greater than or equal to a reference similarity, the security message filtering module (142) determines that the user's input message or the user's attached data contains the content of the security document. If the similarity between the user's input message or the user's attached data and the security document is greater than or equal to the reference similarity, the security message filtering module (142) blocks the transmission of the message so that the user's input message or the attached data is not transmitted to the ChatGPT server (300).

[0086] Additionally, a messenger program executed using the processor (140, 240) can create a secure messenger channel. Furthermore, the secure message filtering module (142) can create snapshots of some or all of the messages occurring in the secure messenger channel. The secure message filtering module (142) can perform security learning based on the snapshots of the secure messenger channel, thereby learning about security regulations and security information and secure message filtering. At this time, the messages included in the snapshots can include not only messages input by the user but also response messages responded to by the ChatGPT server (300).

[0087] In addition, the security message filtering module (142) performs word and context analysis of the message based on the user's input message and / or the response message of the ChatGPT server (300). In addition, the security filtering of the message is performed based on one or more of security regulations, security keywords, security information patterns, and security documents for each security level and the word and context analysis results included in the user's input message. In addition, if the security message filtering module (142) determines that the security level of the user's input message violates the security level set for each user, the transmission of the message is blocked so that the user's input message is not transmitted to the ChatGPT server (300).

[0088] The notification generation module (143) performs the function of generating notification information regarding security reviews of user messages and violations of security regulations and providing notifications to users and administrators.

[0089] Specifically, when security filtering is performed on a user's input message, the notification generation module (143) can generate a message security level violation notification to the user, indicating that the message contains information that violates security regulations, has been filtered, and has been transmitted to the ChatGPT server (300).

[0090] In addition, when security filtering is performed on a user's input message, the notification generation module (143) can generate a notification including information related to the user's input message and the security filtering performed, and provide the notification to the administrator account.

[0091] Hereinafter, a method for improving security using ChatGPT using a messenger program according to an embodiment of the present invention will be described in detail with reference to FIG. 5.

[0092] Figure 5 is a flowchart of a method for improving security using chat GPT using a messenger program according to an embodiment of the present invention.

[0093] Referring to FIG. 5, a method for improving security using ChatGPT using a messenger program includes a security information setting step (S100), a message receiving step (S200), a security message filtering step (S300), a ChatGTP server (300) transmission step (S400), and a ChatGPT server (300) response receiving step (S500).

[0094] In the security information setting step (S100), the messenger server (100) can set security criteria for filtering content requiring security, such as security regulations and information.

[0095] In the security information setting step (S100), the messenger server (100) can set the target of security regulations and set different security levels for each target of security regulations. The messenger server (100) can set security regulations and security information for each security level and set the target of security level application.

[0096] In addition, in the security information setting step (S100), the messenger server (100) can set the target of application of the security regulations, and can set different security levels for each target of application of the security regulations. In the security information setting step (S100), the messenger server (100) can set security management information regarding security filtering criteria, such as security regulations, security keywords, security information patterns, and security documents, and can set different security management information depending on the security level.

[0097] In the security message receiving step (S200), the messenger server (100) receives a message entered by the user using a messenger program executed using the user's terminal (200).

[0098] In the security message filtering step (S300), the messenger server (100) filters whether the message entered by the user contains information that does not comply with security regulations and information.

[0099] In the security message filtering step (S300), the messenger server (100) performs learning to filter whether a message entered by a user complies with security regulations and includes security information based on the security management information set in the security information setting step (S100).

[0100] In the security message filtering step (S300), the messenger server (100) can perform natural language processing using a language model such as a Large Language Model (LLM). In addition, the messenger server (100) repeatedly performs pre-training and fine-tuning on a data set of security management information based on a BERT (Bidirectional Encoder Representations from Transformers) model, thereby continuously learning whether the user's input message complies with security regulations and contains security information.

[0101] In the secure message filtering step (S300), the messenger server (100) can convert the user's input message filtering results into a learnable dataset and perform learning using a BERT (Bidirectional Encoder Representations from Transformers) model. The structure of the messenger server (100) is not limited to the BERT model, and can perform keyword-by-keyword classification of fan messages by inputting it into a deep learning structure such as a recurrent neural network (RNN) or a convolutional neural network (CNN).

[0102] When the messenger server (100) uses a convolutional neural network (CNN) structure, the convolutional neural network (CNN) structure may be formed to include a sentence matrix, a convolutional layer, a pooling layer, and a softmax function.

[0103] In the security message filtering step (S300), the messenger server (100) performs filtering on the user's input message based on security regulations by security level, question-and-answer monitoring evaluation, security keywords, security information patterns, and question-and-answer filtering results, thereby preventing information requiring security from being leaked to the chat GPT server (300).

[0104] In the security message filtering step (S300), the messenger server (100) can perform segmentation of security regulations based on the user's input message and the response information of the chat GPT server (300) to the input message.

[0105] The messenger server (100) can derive a level of security compliance by evaluating whether a message entered by a user and a response from the ChatGPT server comply with security regulations.

[0106] Additionally, in the secure message filtering step (S300), the messenger server (100) may provide a monitoring interface that displays information about users' input messages so that an administrator can monitor the users' input messages. The monitoring interface may include an area that displays the validity and security compliance of the message filtering results of the messenger server (100).

[0107] In addition, in the security message filtering step (S300), when filtering is performed on a user's input message, the messenger server (100) can generate and provide a filtering information provision interface that provides the user with information on the filtered content and the filtering reason.

[0108] Additionally, in the secure message filtering step (S300), the messenger server (100) determines whether the user's input message contains a security keyword, and if the user's input message contains a security keyword, the messenger server (100) may replace the security keyword with a substitute word or delete the security keyword. In addition, the messenger server (100) may perform filtering to prevent the user's input message containing the security keyword from being transmitted to the ChatGPT server (300).

[0109] Specifically, if the user's input message includes a first security keyword, the messenger server (100) can replace the first security keyword with a substitute word. Furthermore, if the user's input message includes a second security keyword, the messenger server (100) blocks the transmission of the user's input message so that the user's input message containing the second security keyword is not transmitted to the ChatGPT server (300).

[0110] Accordingly, the messenger server (100) performs conversion to an alternative keyword in the case of a keyword with a low security level, but can prevent the message itself from being transmitted to the ChatGPT server (300) in the case of a keyword with a high security level. In this case, the second security keyword may refer to a keyword set to correspond to a security level higher than that of the first security keyword.

[0111] In addition, in the security message filtering step (S300), if the user's input message includes characters corresponding to a security information pattern, the messenger server (100) can replace the characters corresponding to the security information pattern with alternative characters and transmit the user's input message to the chatGPT server (300).

[0112] At this time, the security information pattern means information having certain rules and patterns, such as a phone number, resident registration number, address, password, etc., and the messenger server (100) can perform filtering of security information having such patterns to prevent information requiring security from being transmitted to the chat GPT server (300).

[0113] Additionally, in the secure message filtering step (S300), the messenger server (100) derives the similarity between the user's input message or the user's attached data and the secure document. If the derived similarity is higher than the standard similarity, it is determined that the user's input message or the user's attached data contains the content of the secure document. If the similarity between the user's input message or the user's attached data and the secure document is higher than the standard similarity, the messenger server (100) blocks the transmission of the message so that the user's input message or the attached data is not transmitted to the ChatGPT server (300).

[0114] Additionally, a messenger program executed using a processor (140, 240) can create a secure messenger channel. In the secure message filtering step (S300), the messenger server (100) can create snapshots of some or all of the messages generated in the secure messenger channel. The messenger server (100) can perform security learning based on the snapshots of the secure messenger channel, thereby learning about security regulations and security information and secure message filtering. At this time, the messages included in the snapshots can include not only messages input by the user but also response messages responded to by the ChatGPT server (300).

[0115] In addition, in the security message filtering step (S300), the messenger server (100) performs word and context analysis of the message based on the user's input message and / or the response message of the ChatGPT server (300). Then, the message is security filtered based on one or more of security regulations, security keywords, security information patterns, and security documents for each security level and the word and context analysis results included in the user's input message. Then, if the messenger server (100) determines that the security level of the user's input message violates the security level set for each user, the messenger server blocks the transmission of the message so that the user's input message is not transmitted to the ChatGPT server (300).

[0116] In the Chat GPT server transmission step (S400), the messenger server (100) transmits the user's input message, for which filtering has been completed, to the Chat GPT server (300). In the Chat GPT server transmission step (S400), the messenger server (100) changes the security characters and security keywords included in the user's input message into replacement characters and replacement keywords and transmits them to the Chat GPT server (300).

[0117] In the Chat GPT server response reception step (S500), the Chat GPT server (300) can generate a response message or response data for the message received from the messenger server (100) in the Chat GPT server transmission step (S400) and transmit the same to the messenger server (100). Then, the messenger server (100) provides the response message received from the Chat GPT server (300) to the user terminal (200).

[0118] Accordingly, the messenger server (100) can perform learning on security regulations and information to prevent security information from being leaked to the chat GPT server (300).

[0119] The foregoing description of the present invention is for illustrative purposes only, and those skilled in the art will readily appreciate that the present invention can be readily modified into other specific forms without altering the technical spirit or essential characteristics of the present invention. Therefore, the embodiments described above should be understood as illustrative in all respects and not restrictive. For example, each component described as a single entity may be implemented in a distributed manner, and similarly, components described as distributed may be implemented in a combined manner.

[0120] The scope of the present invention is indicated by the claims described below rather than the detailed description above, and all changes or modifications derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present invention.

[0121] The form for carrying out the invention is the same as the best form for carrying out the invention described above.

[0122] The present invention relates to a method and device for improving security using ChatGPT using a messenger program, and is applicable to industries related to artificial intelligence and messengers, and thus has industrial applicability.

Claims

1. How to use a messenger program linked with ChatGPT: Steps to set up the messenger program's security rules and security information, When a user inputs a message, a step of performing security filtering of the message using a messenger program learned based on the security regulations and the security information; A step of transmitting the message for which the above security filtering has been completed to the ChatGPT server, and A step of receiving a response to the message from the ChatGPT server and providing it to the user. A method for improving security using ChatGPT using a messenger program, including:

2. In paragraph 1, The steps for setting the above security regulations and security information are: A step for setting the security level of the above security regulations and the above security information, and Steps to set the target for application by security level A method for improving security using ChatGPT using a messenger program, including:

3. In paragraph 2, The steps for setting the above security regulations and security information are: Steps to set information on one or more of the following: security regulations, security keywords, security information patterns, and security documents for each security level. A method for improving security using ChatGPT using a messenger program, which further includes:

4. In paragraph 3, The steps for performing security filtering of the above message are: A step of performing context and word learning on at least one of the security regulations, security keywords, security information patterns, and security documents for each security level using the messenger program learned using BERT (Bidirectional Encoder Representations from Transformers), and A step of performing security filtering on the message based on the above learning results. A method for improving security using ChatGPT using a messenger program, including:

5. In paragraph 4, The steps for performing security filtering of the above message are: If the above message includes a first security keyword, a step of replacing the first security keyword with a replacement word, and If the above message contains a second security keyword, a step of blocking the transmission of the message so that the message is not transmitted to the ChatGPT server. A method for improving security using ChatGPT using a messenger program, which further includes:

6. In paragraph 5, The steps for performing security filtering of the above message are: If the message above includes a character corresponding to the security information pattern, a step of replacing the character corresponding to the security information pattern with a replacement character. A method for improving security using ChatGPT using a messenger program, which further includes:

7. In paragraph 6, The steps for performing security filtering of the above message are: A step of deriving a first similarity between the above message and the above security document, and If the first similarity is greater than or equal to the first reference similarity, a step of blocking transmission of the message so that the message is not transmitted to the ChatGPT server. A method for improving security using ChatGPT using a messenger program, which further includes:

8. In paragraph 7, The steps for performing security filtering of the above message are: Steps to create a secure messenger channel: A step of creating a snapshot of some or all of the messages occurring in the above secure messenger channel, and A step of performing security learning of the messenger program based on the above snapshot. A method for improving security using ChatGPT using a messenger program, which further includes:

9. In paragraph 8, The steps for performing security filtering of the above message are: A step of performing word and context analysis of the above message, and A step of performing security filtering on a message based on one or more of the security regulations, security keywords, security information patterns, and security documents for each security level and the results of word and context analysis included in the message, and blocking the transmission of the message so that the message is not transmitted to the ChatGPT server if the security level of the message is determined to be a message that violates the security level set for each user. A method for improving security using ChatGPT using a messenger program, which further includes:

10. In paragraph 9, The steps for performing security filtering of the above message are: When security filtering of the above message is performed, a step of generating and displaying a notification including information that security filtering has been performed on the message because the message contains information that violates security regulations, and When the above security filtering is performed, a step of generating an administrator notification including the message and information on the security filtering performed and providing the administrator notification to the administrator account. A method for improving security using ChatGPT using a messenger program, which further includes:

11. A device that provides a message service using a messenger server linked to a ChatGPT server. A communication module that performs information transmission and reception between the terminal and the messenger server and the chatGPT server, A memory that stores at least one instruction, and A processor that executes at least one of the above instructions Includes, The processor, by executing the at least one instruction, Set the security rules and security information of the messenger program, When a user inputs a message, security filtering of the message is performed using a messenger program learned based on the above security regulations and the above security information. The message for which the above security filtering has been completed is transmitted to the ChatGPT server. A device for enhancing security using ChatGPT by using a messenger program that receives a response to the message from the ChatGPT server and provides it to the user.

12. In paragraph 11, The above processor is a security enhancement device using ChatGPT using a messenger program that sets the security level of the above security regulations and the above security information and sets the application target for each security level.

13. In paragraph 12, The above processor is a security enhancement device using ChatGPT, which uses a messenger program to set information on one or more of security regulations, security keywords, security information patterns, and security documents by security level.

14. In paragraph 13, The above processor performs context and word learning on at least one of security regulations, security keywords, security information patterns, and security documents by security level using the messenger program learned using BERT (Bidirectional Encoder Representations from Transformers), A security enhancement device using ChatGPT, a messenger program, which performs security filtering on the message based on the above learning results.

15. In paragraph 14, A ChatGPT security enhancement device using a messenger program, wherein the processor, when the message includes a first security keyword, replaces the first security keyword with a substitute word, and when the message includes a second security keyword, blocks transmission of the message so that the message is not transmitted to the ChatGPT server.

16. In paragraph 15, The above processor is a security enhancement device using ChatGPT using a messenger program, which replaces the characters corresponding to the security information pattern with alternative characters when the message includes characters corresponding to the security information pattern.

17. In paragraph 16, A ChatGPT security enhancement device using a messenger program, wherein the processor derives a first similarity between the message and the security document, and if the first similarity is greater than or equal to a first reference similarity, blocks transmission of the message so that the message is not transmitted to the ChatGPT server.

18. In paragraph 17, A security enhancement device using ChatGPT using a messenger program, wherein the processor creates a secure messenger channel, creates a snapshot of some or all of the messages occurring in the secure messenger channel, and performs security learning of the messenger program based on the snapshot.

19. In paragraph 18, A ChatGPT security enhancement device using a messenger program, wherein the processor performs word and context analysis of the message, and performs security filtering of the message based on the results of word and context analysis included in the message and one or more of the security regulations, security keywords, security information patterns, and security documents for each security level, and if the security level of the message is determined to be a message that violates the security level set for each user, the transmission of the message is blocked so that the message is not transmitted to the ChatGPT server.

20. In paragraph 19, The above processor, when security filtering of the message is performed, generates a notification including information that the message contains information that violates security regulations and that filtering of the message has been performed, and displays the notification to the user, and when the security filtering is performed, generates an administrator notification including the message and information on the security filtering performed, and provides the notification to the administrator account. A security enhancement device using a messenger program, ChatGPT.

Images