An AI object-oriented sensitive information isolation processing method and device

Abstract

The application discloses an AI object-oriented sensitive information isolation processing method and device, wherein the sensitive information stored by a user is stored in a framework system, and the sensitive information is divided into declaration data and content data; the declaration data is provided to an AI object, so that the AI object can freely read; the content data is maintained by the framework system, so that the AI object cannot directly read; the AI object executes an inference task according to the declaration data and a demand proposed by the user, generates return content, and sends the return content to the framework system; and the framework system performs a setting processing operation according to a setting rule according to an indication in the return content, and obtains a processing result. The application can ensure that the AI object cannot directly contact the core content of the sensitive information, and can efficiently and accurately complete task planning.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, in particular to a sensitive information isolation processing method and device for AI objects. BACKGROUND

[0002] With the wide application of artificial intelligence technology, the scene of users using AI assistants to plan and execute complex tasks is increasing. In this process, AI often needs to access sensitive information such as ID card numbers and mobile phone numbers. However, directly exposing these sensitive information to AI models may cause data leakage risks and other security problems.

[0003] Existing data desensitization, encryption processing, and limiting AI direct access to sensitive information solutions have the defects of information restoration difficulty or inability to completely prevent AI from contacting sensitive information. For example, although MCP (Model Context Protocol) requires user authorization for AI to access sensitive data, AI can still read the content of sensitive data after obtaining authorization, which makes it difficult to fundamentally guarantee the security of sensitive information.

[0004] Therefore, how to invent a sensitive information isolation processing method that can isolate sensitive information when AI assistants reason and ensure that AI cannot directly contact the core content of sensitive information, while efficiently and accurately completing task planning, has become a problem to be solved. SUMMARY

[0005] Therefore, the present application provides a sensitive information isolation processing method and device for AI objects, which divides sensitive information into declaration information and sensitive content, manages complete information within the framework, and only exposes declaration information to AI objects. AI objects reason based on declaration information and tasks without contacting sensitive information content. The reasoning result indicates how the framework processes the subsequent processing, and the indication includes how to use sensitive information content. Since the subsequent processing is performed by the framework, it ensures that AI objects have not contacted sensitive information content from the beginning to the end, while efficiently and accurately completing task planning.

[0006] In order to achieve the above purpose, the present application provides the following technical scheme: a sensitive information isolation processing method for AI objects, comprising:

[0007] storing the sensitive information stored by the user in the framework system, and dividing the sensitive information into declaration data and content data; providing the declaration data to the AI object to allow the AI object to freely read; and maintaining the content data by the framework system to prevent the AI object from directly reading;

[0008] The AI object executes an inference task according to the declaration data and a demand raised by a user, generates return content, and sends the return content to the framework system.

[0009] The framework system performs a setting processing operation according to a setting rule according to an indication in the return content, and obtains a processing result.

[0010] As a preferred scheme of the sensitive information isolation processing method for the AI object, the framework system and the AI object perform data interaction through a setting data protocol; the setting data protocol includes: a meaning and a format of the declaration data received by the AI object; a representation manner of an indication of the framework system using sensitive information in the return content sent by the AI object; and an operation to be performed according to the indication.

[0011] As a preferred scheme of the sensitive information isolation processing method for the AI object, the declaration data includes: a variable name, a variable type, and content description of a variable; a file name, a format, a size, file content description, and file use description of a file; and system version, environment variable, software version, and system setting description information in a running environment setting.

[0012] The content data includes: a variable value of a variable; a file itself of a file; and a data file required for constructing a corresponding running environment in a running environment setting.

[0013] As a preferred scheme of the sensitive information isolation processing method for the AI object, in a process in which the AI object executes an inference task according to the declaration data and a demand raised by a user, the AI object can read all the declaration data, and a reading manner includes: one is that a variable declaration data list is spliced into a prompt before the framework system sends the prompt to the AI object; and two is that the AI object autonomously reads the declaration data.

[0014] As a preferred scheme of the sensitive information isolation processing method for the AI object, in a process in which the framework system performs a setting processing operation according to a setting rule according to an indication in the return content, for a variable reference identifier: the framework system replaces the variable reference identifier with a variable value or inserts a corresponding variable assignment statement in an instruction according to a variable name in the variable reference identifier.

[0015] For a running environment setting indication: the framework system sends the running environment setting indication to a command execution module.

[0016] For an execution command indication: the framework system sends the execution command indication to the command execution module.

[0017] As a preferred scheme of the sensitive information isolation processing method for the AI object, the command execution module executes the command indication sent from the framework system, and in the process of executing the corresponding command, the command execution module:

[0018] When the running environment setting indication is received, the command execution module constructs a required running environment according to the parameters indicated in the running environment setting indication; and when the execution command indication is received, the command execution module executes the command in the constructed running environment according to the execution command indication.

[0019] The application further provides a sensitive information isolation processing device for an AI object, based on the sensitive information isolation processing method for the AI object, comprising:

[0020] The sensitive information classification and maintenance unit is used for storing the sensitive information stored by the user in the framework system, and classifying the sensitive information into declaration data and content data; the declaration data is provided to the AI object, so that the AI object can freely read; and the content data is maintained by the framework system, so that the AI object cannot directly read;

[0021] The AI object reasoning task execution unit is used for the AI object to execute a reasoning task according to the declaration data and the demand proposed by the user, to generate return content, and to send the return content to the framework system;

[0022] The framework system processing unit is used for the framework system to perform a setting processing operation according to the indication in the return content, to obtain a processing result.

[0023] As a preferred scheme of the sensitive information isolation processing device for the AI object, in the AI object reasoning task execution unit and the framework system processing unit, the framework system and the AI object perform data interaction through a setting data protocol; the setting data protocol comprises: the meaning and format of the declaration data received by the AI object; the representation manner of the return content sent by the AI object, which indicates that the framework system uses sensitive information; and the operation required to be executed by the framework system according to the indication.

[0024] As a preferred scheme of the sensitive information isolation processing device for the AI object, in the sensitive information classification and maintenance unit, the declaration data comprises: a variable name, a variable type and content description of a variable; a file name, a format, a size, a file content description and a file use description of a file; system version, environment variable, software version and system setting description information in a running environment setting;

[0025] The content data includes variable values of variables, files themselves of files, and data files required for building a corresponding running environment in a running environment setting.

[0026] As a preferred scheme of the sensitive information isolation processing device for AI objects, in the AI object reasoning task execution unit, in the process of the AI object executing a reasoning task according to the declaration data and the demand proposed by the user, the AI object can read all the declaration data, and the reading mode includes: 1. before the framework system sends the Prompt to the AI object, splicing the variable declaration data list into the Prompt; 2. the AI object independently reads the declaration data.

[0027] As a preferred scheme of the sensitive information isolation processing device for AI objects, in the framework system processing unit, in the process of the framework system performing the setting processing operation according to the indication in the returned content and according to the setting rule, for the variable reference identifier: the framework system replaces the variable reference identifier with a variable value or inserts a corresponding variable assignment statement in the instruction according to the variable name in the variable reference identifier.

[0028] For the running environment setting indication: the framework system sends the running environment setting indication to the command execution module.

[0029] For the execution command indication: the framework system sends the execution command indication to the command execution module.

[0030] As a preferred scheme of the sensitive information isolation processing device for AI objects, in the processing result display unit, the command execution module executes the command indication sent from the framework system, and in the process of the command execution module executing the corresponding command:

[0031] When receiving the running environment setting indication, the command execution module builds a running environment meeting the requirements according to the parameters indicated in the running environment setting indication; when receiving the execution command indication, the command execution module executes the command in the built running environment according to the execution command indication.

[0032] The application has the following advantages: the application stores the sensitive information stored by the user in the framework system, and divides the sensitive information into declaration data and content data; the declaration data is provided to the AI object, so that the AI object can read freely; the content data is maintained by the framework system, so that the AI object cannot read directly; the AI object performs an inference task according to the declaration data and the demand proposed by the user, generates return content, and sends the return content to the framework system; and the framework system performs a setting processing operation according to the indication in the return content, and obtains a processing result according to the setting rules. The application can guarantee the safety of sensitive information, divide the sensitive information into declaration data and content data, expose only the declaration data to the AI object, and manage the complete information by the framework. The framework ensures the safety of sensitive information in the storage, transmission and use process through encryption technology and access control mechanism. The AI object does not contact the sensitive information content from the beginning to the end, and the risk of data leakage is avoided, for example, the sensitive data such as the personal information and the mailbox password of the user in the case are effectively protected. The application improves the task execution efficiency and accuracy. The AI object can read the declaration data to perform inference, generate return content according to the demand of the user, instruct the framework to perform subsequent operations, and the framework performs corresponding operations such as variable replacement, environment construction and command execution according to the indication, so that the task can be completed efficiently and accurately. The application has strong flexibility and adaptability, and has good scalability, and meets the changing business needs. BRIEF DESCRIPTION OF DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the application or the technical solutions in the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only exemplary, and for those skilled in the art, other drawings can be derived from the provided drawings without creative labor.

[0034] The structures, proportions, sizes, etc. shown in the specification are only used to cooperate with the content disclosed in the specification, to be understood and read by those skilled in the art, and are not used to limit the limiting conditions that the application can be implemented, so they do not have technical significance. Any modification of structure, change of proportion relationship or adjustment of size, without affecting the effect and purpose that the application can produce, should still fall within the scope covered by the disclosed technical content.

[0035] Figure 1 A flowchart of a sensitive information isolation processing method for an AI object provided in embodiment 1 of the application;

[0036] Figure 2 A specific implementation diagram of a sensitive information isolation processing method for an AI object provided in embodiment 1 of the application;

[0037] Figure 3 A schematic diagram of an AI object-oriented sensitive information isolation processing device architecture provided in Embodiment 2 of the present application. DETAILED DESCRIPTION

[0038] The embodiments of the present application will be described in detail by specific embodiments, and those skilled in the art can easily understand other advantages and effects of the present application from the disclosed content. Obviously, the described embodiments are part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0039] Embodiment 1

[0040] Reference Figure 1 and Figure 2 , Embodiment 1 of the present application provides an AI object-oriented sensitive information isolation processing method, comprising the following steps:

[0041] S1, store the sensitive information stored by the user in the framework system, and divide the sensitive information into declaration data and content data; provide the declaration data to the AI object, so that the AI object can be freely read; maintain the content data through the framework system, so that the AI object cannot directly read;

[0042] S2, the AI object executes the reasoning task according to the declaration data and the demand put forward by the user, generates the return content, and sends the return content to the framework system;

[0043] S3, the framework system performs setting processing operation according to the indication in the return content, and obtains the processing result according to the setting rule.

[0044] In this embodiment, in step S1, the sensitive information stored by the user is stored in the framework system, and the sensitive information is divided into declaration data and content data; the declaration data is provided to the AI object, so that the AI object can be freely read; the content data is maintained through the framework system, so that the AI object cannot directly read;

[0045] Specifically, when the user stores the sensitive information in the framework system, the framework system first starts the access verification program. Verify the legality of user identity and the integrity of information transmission, ensure the reliable source and non-tampering of information through digital signature, identity token and other technologies. Then, according to the pre-defined data format specification, the sensitive information is preliminarily sorted and converted into a standard format that can be recognized and processed by the system, to prepare for subsequent classification.

[0046] The framework system uses natural language processing techniques and pre-defined sensitive information classification rules to conduct in-depth analysis on the sorted information. If the information is in text form, the framework system will identify possible sensitive words and phrases, such as keywords related to personal identity, finance, health, etc. According to the identification results, the sensitive information is accurately classified into variables, files, operating environment settings, etc. For information that is difficult to identify, the framework system will start an artificial review process, and professional personnel will assist in determining the category.

[0047] For different categories of sensitive information, the framework system performs corresponding separation operations. For variable-type sensitive information, the variable name, variable type, content description, etc. are extracted as declaration data, and the variable value is taken as content data; for file-type sensitive information, the file name, format, size, file content description, file usage description, etc. constitute the declaration data, and the file itself is the content data; for operating environment setting-type sensitive information, system version, environment variable, software version, system setting, etc. are the declaration data, and the data file required to build the corresponding operating environment is taken as the content data. During the separation process, the system generates a unique identifier for each information segment, which facilitates subsequent management and calling.

[0048] The framework system plans different storage strategies for declaration data and content data. The declaration data is stored in a high-performance distributed cache server to allow AI objects to quickly read. The cache server sets access frequency limits and data expiration mechanisms to prevent malicious reading and long-term exposure of data risks. The content data is stored in a special storage device with multiple encryption and access control. AES encryption algorithm is used to encrypt the content data, and in terms of access control, only authorized specific modules within the framework system can access and process the content data under strict audit supervision.

[0049] In this embodiment, the framework system ensures the security of sensitive information during storage, transmission, and use through encryption technology and access control mechanisms. The framework system can be located on the server or on the client. The framework system located on the client can prove its security through open source, packet capture, etc. Sensitive information can be stored in the framework system or in other systems and read interfaces can be developed to the framework.

[0050] In this embodiment, the variable storage structure based on the JSON data structure is as follows:

[0051]

[0052]

[0053] Among them, the variable, type, describe fields are declaration data, which will be visible to AI objects, and the value field is content data, which is invisible to AI objects.

[0054] In this embodiment, the file information storage structure based on the JSON data structure is as follows:

[0055]

[0056] Among them, file_name, id, format, size, description are declaration data visible to AI objects, and file content is stored in the path path of the client. The path field and file content are invisible to AI objects.

[0057] In this embodiment, the running environment information storage structure based on the JSON data structure is as follows:

[0058]

[0059]

[0060]

[0061] Among them, all fields are mirror information, which is visible to AI objects, and the mirror file itself is invisible to AI objects.

[0062] In this embodiment, in step S2, the AI object executes the inference task according to the declaration data and the demand put forward by the user, generates the return content, and sends the return content to the framework system;

[0063] Specifically, when the AI object executes the inference task, it can read all the declaration data, and the reading method is flexible. It can either splice the variable declaration data list into the prompt before the framework sends the prompt to the AI object, or allow the AI object to read the declaration data independently during the inference process. The AI object combines the specific demand put forward by the user with these declaration data to perform task inference and generate return content sent to the framework. The return content clearly indicates the operation direction of the framework using sensitive information in the subsequent operation, including variable filling, running environment preparation, command execution and other key links, and provides guidance for subsequent processing.

[0064] In this embodiment, the framework system and the AI object interact through a set data protocol; the set data protocol includes: the meaning and format of the declaration data received by the AI object; the representation method of the return content sent by the AI object indicating that the framework system uses sensitive information; and the operation that the framework system needs to perform according to the indication.

[0065] In this embodiment, in step S3, the framework system performs setting processing operation according to the indication in the return content according to the setting rules, and obtains a processing result;

[0066] Specifically, for a variable reference identifier, the framework system replaces the variable reference identifier with a variable value according to the variable name in the variable reference identifier, or inserts a corresponding variable assignment statement in the instruction;

[0067] For a running environment setting indication, the framework system sends the running environment setting indication to the command execution module;

[0068] For an execution command indication, the framework system sends the execution command indication to the command execution module.

[0069] In a possible embodiment, after step S, the AI object displays the processing result to the user.

[0070] Specifically, if the return content is a variable reference identifier, the AI object displays the variable value replaced by the framework system to the user;

[0071] If the return content is a running environment setting indication or an execution command indication, the framework system sends the command indication to the command execution module for execution; the command execution module executes the command indication sent from the framework system, and feeds back the execution process and result to the AI object; the AI object displays the execution process and result to the user;

[0072] During the execution of the corresponding command by the command execution module:

[0073] When receiving the running environment setting indication, the command execution module constructs a running environment meeting the requirements according to the parameters indicated in the running environment setting indication; when receiving the execution command indication, the command execution module executes the command in the constructed running environment according to the execution command indication.

[0074] In a possible embodiment, the command execution module can be located on the server side or the client side, and can be a component in the framework system or an external module accessed to the framework through a protocol. The framework system and the command execution module can perform security check before the execution of the instruction, to avoid the execution of unsafe instructions generated by the AI object. The running environment provided by the command execution module should meet the settings in the return result of the AI object, so that the command returned by the AI object can be executed in the expected environment.

[0075] In a possible embodiment, a specific sensitive information isolation example is provided as follows:

[0076] T1, the user stores the user's personal information in the form of variables in the client:

[0077]

[0078]

[0079] At the same time, the user hosts the mailbox configuration file in the client, and the declaration information is as follows:

[0080]

[0081] The mailbox configuration file content is as follows:

[0082]

[0083]

[0084] T2, the user sends a task through the client: send my address to recipient@example.com by email;

[0085] T3, the client forms a prompt by combining the declaration data and user requirements:

[0086]

[0087]

[0088] Available file declaration:

[0089] Available operating environment declaration:

[0090]

[0091]

[0092] Question: send my address to recipient@example.com by email;

[0093] T4, the AI object generates a return result according to the above prompt:

[0094]

[0095]

[0096] T5, the framework receives the above content and parses it, and according to the content of the task field, it knows that the operation to be done later is to execute the command; then parse the command field, replace the variables referred to in it with actual values, and get the actual command to be executed:

[0097] "echo" name: Wang Xiaoming\naddress: Beijing Haidian District A Community X Building 123 Room\nphone: 123456789 "| msmtp recipient@example.com";

[0098] T6, prepare the running environment of the command, parse the content under environment:

[0099] "Determine the container image to be used according to image_id, according to the files field, the file with id 12345565 needs to be placed on the ~ / .msmtprc path of the running environment, and the running environment is constructed according to this requirement"

[0100] T7, after the running environment is constructed, the command is sent to the running environment for execution;

[0101] T8, finally, the whole process and result are shown to the user.

[0102] In summary, the present application stores the sensitive information stored by the user in the framework system, and divides the sensitive information into declaration data and content data; the declaration data is provided to the AI object, so that the AI object can be freely read; the content data is maintained by the framework system, so that the AI object cannot directly read; the AI object executes the reasoning task according to the declaration data and the demand put forward by the user, generates the return content, and sends the return content to the framework system; the framework system performs setting processing operation according to the indication in the return content, and obtains the processing result. The present application can guarantee the safety of sensitive information, by dividing the sensitive information into declaration data and content data, only exposing the declaration data to the AI object, and the complete information is managed by the framework, and the framework ensures the safety of sensitive information in the storage, transmission and use process through encryption technology and access control mechanism. The AI object has not contacted the sensitive information content from the beginning to the end, avoiding the risk of data leakage, such as the effective protection of sensitive data such as user personal information and mailbox password in the case. The present application improves the task execution efficiency and accuracy, the AI object can read the declaration data for reasoning, generates the return content according to the user demand, and instructs the framework to perform subsequent operation, and the framework performs corresponding operation according to the indication, such as variable replacement, environment construction and command execution, so that the task can be efficiently and accurately completed. The present application has strong flexibility and adaptability, and has good scalability, and meets the changing business needs.

[0103] It should be noted that the method of the embodiments of the present disclosure can be executed by a single device, such as a computer or a server, etc. The method of the embodiments can also be applied to a distributed scenario, and be completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only execute one or more steps in the method of the embodiments of the present disclosure, and the multiple devices can interact with each other to complete the method.

[0104] It should be noted that some embodiments of the present disclosure are described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in a different order and still achieve desirable results. Additionally, the processes depicted in the figures do not necessarily require the particular order shown, or sequential order, to achieve the desired results. In certain implementations, multitasking and parallel processing can be advantageous.

[0105] Embodiment 2

[0106] Referring to Figure 3 Embodiment 2 of the present disclosure also provides an AI object-oriented sensitive information isolation processing device, comprising:

[0107] A sensitive information classification and maintenance unit 001 is configured to store the sensitive information stored by a user in a framework system, and divide the sensitive information into declaration data and content data; provide the declaration data to an AI object, so that the AI object can freely read; and maintain the content data through the framework system, so that the AI object cannot directly read;

[0108] An AI object reasoning task execution unit 002 is configured to execute a reasoning task according to the declaration data and a demand proposed by a user, generate a return content, and send the return content to the framework system;

[0109] A framework system processing unit 003 is configured to perform a set processing operation according to a set rule, and obtain a processing result, according to an indication in the return content.

[0110] In the present embodiment, the AI object reasoning task execution unit 002 and the framework system processing unit 003 perform data interaction between the framework system and the AI object through a set data protocol; the set data protocol includes: the meaning and format of the declaration data received by the AI object; the representation of the return content sent by the AI object, which indicates the use of sensitive information by the framework system; and the operation to be performed by the framework system according to the indication.

[0111] In this embodiment, the sensitive information classification and maintenance unit 001, the declaration data includes: variable name, variable type, content description of variable; file name, format, size, file content description, file usage description of file; system version, environment variable, software version, system setting description information in running environment setting;

[0112] The content data includes: variable value of variable; file itself of file; data file required for building corresponding running environment in running environment setting.

[0113] In this embodiment, the AI object reasoning task execution unit 002, in the process of AI object executing reasoning task according to the declaration data and the demand put forward by the user, the AI object can read all the declaration data, and the reading mode includes: one is to splice the variable declaration data list into the prompt before the framework system sends the prompt to the AI object; two is that the AI object reads the declaration data independently.

[0114] In this embodiment, the framework system processing unit 003, in the process of the framework system performing the setting processing operation according to the setting rules according to the indication in the returned content, for variable reference identifier: the framework system replaces the variable reference identifier with variable value or inserts corresponding variable assignment statement in instruction according to variable name in the variable reference identifier;

[0115] For running environment setting indication: the framework system sends the running environment setting indication to the command execution module;

[0116] For execution command indication: the framework system sends the execution command indication to the command execution module.

[0117] In a possible embodiment, it further includes a processing result display unit 004, in the processing result display unit 004, the command execution module executes the command indication sent from the framework system, and feeds back the execution process and result to the AI object; the AI object displays the execution process and result to the user;

[0118] In the process of the command execution module executing corresponding command:

[0119] When receiving the running environment setting indication, the command execution module constructs the running environment meeting the requirements according to the parameters indicated in the running environment setting indication; when receiving the execution command indication, the command execution module executes the command in the constructed running environment according to the execution command indication.

[0120] It is to be explained that the information interaction and execution process between the modules of the system described above, since the same concept as the method embodiment in the embodiment 1 of the present application, the technical effects brought by it are the same as the method embodiment of the present application, and the specific content can be referred to the description in the method embodiment of the present application, which will not be repeated here.

[0121] Embodiment 3

[0122] The embodiment 3 of the present application provides a non-transitory computer readable storage medium, which stores a program code of a sensitive information isolation processing method for AI object, and the program code includes instructions for executing the sensitive information isolation processing method for AI object of the embodiment 1 or any possible implementation manner thereof.

[0123] The computer readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. integrated with one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk (SSD)) and the like.

[0124] Embodiment 4

[0125] The embodiment 4 of the present application provides an electronic device, which includes a memory and a processor.

[0126] The processor and the memory complete the communication between each other through a bus; the memory stores program instructions that can be executed by the processor, and the processor calling the program instructions can execute the sensitive information isolation processing method for AI object of the embodiment 1 or any possible implementation manner thereof.

[0127] Specifically, the processor can be implemented by hardware or software, when implemented by hardware, the processor can be a logic circuit, an integrated circuit, etc.; when implemented by software, the processor can be a general-purpose processor, which is implemented by reading the software code stored in the memory, and the memory can be integrated in the processor or located outside the processor and exist independently.

[0128] In the embodiments described above, all or some of the modules / units can be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or some of the modules / units can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded into and executed by a computer, all or some of the procedures or functions as described in the embodiments of the present application are generated. The computer can be a general purpose computer, a special purpose computer, a computer network, or other programmable system. The computer instructions can be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another computer readable storage medium, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through a wired (for example, coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (for example, infrared, wireless, microwave, etc.) manner.

[0129] It is obvious that those skilled in the art should understand that the modules or steps of the present application described above can be implemented by a general computing system, which can be concentrated on a single computing system or distributed on a network composed of multiple computing systems, and optionally, they can be implemented by program codes executable by a computing system, so that they can be stored in a storage system and executed by a computing system, and in some cases, the steps shown or described can be executed in different order, or they can be made into individual integrated circuit modules, or multiple modules or steps can be made into a single integrated circuit module. Thus, the present application is not limited to any particular combination of hardware and software.

[0130] Although the present application has been described in detail by the above general description and specific embodiments, some modifications or improvements can be made on the basis of the present application, which is obvious to those skilled in the art. Therefore, these modifications or improvements made on the basis of not deviating from the spirit of the present application are within the scope of the present application.

Claims

1. A method for isolating and processing sensitive information of AI objects, characterized in that, include: The sensitive information stored by the user is stored in the framework system, and the sensitive information is divided into declaration data and content data; The declared data is provided to the AI ​​object, allowing the AI ​​object to read it freely; the content data is maintained through the framework system, preventing the AI ​​object from reading it directly. The AI ​​object performs inference tasks based on the declared data and the user's requirements, generates return content, and sends the return content to the framework system. The framework system performs set processing operations according to the instructions in the returned content and the set rules, and obtains the processing results; During the process of the AI ​​object performing inference tasks based on the declared data and the user's requirements: The AI ​​object reads all the declared data in two ways: first, before the framework system sends the Prompt to the AI ​​object, the list of variable declaration data is appended to the Prompt; second, the AI ​​object reads the declared data autonomously. During the process of the framework system performing the setting processing operation according to the instructions in the returned content and the set rules: For variable reference identifiers: The framework system replaces the variable reference identifier with the variable value based on the variable name in the variable reference identifier, or inserts the corresponding variable assignment statement in the instruction; Regarding the runtime environment setting instruction: the framework system sends the runtime environment setting instruction to the command execution module; Regarding the execution command instruction: the framework system sends the execution command instruction to the command execution module; The command execution module executes the command instructions sent from the framework system. During the execution of the corresponding command by the command execution module: When the runtime environment setting instruction is received, the command execution module constructs a runtime environment that meets the requirements according to the parameters specified in the runtime environment setting instruction; when the execution command instruction is received, the command execution module executes the command in the constructed runtime environment according to the execution command instruction.

2. The method for isolating and processing sensitive information of AI objects according to claim 1, characterized in that, The framework system and the AI ​​object interact with each other through a defined data protocol. The defined data protocol includes: the meaning and format of the declared data received by the AI ​​object; the representation method of sensitive information used by the framework system in the returned content sent by the AI ​​object; and the operations that the framework system needs to perform according to the instructions.

3. The method for isolating and processing sensitive information of AI objects according to claim 2, characterized in that, The declared data includes: variable name, variable type, and content description; file name, format, size, file content description, and file usage instructions; system version, environment variables, software version, and system setting information in the runtime environment settings. The content data includes: the variable values; the file itself; and the data files required to build the corresponding runtime environment in the runtime environment settings.

4. A sensitive information isolation and processing device for AI objects, employing the sensitive information isolation and processing method for AI objects as described in any one of claims 1-3, characterized in that, include: The sensitive information classification and maintenance unit is used to store the sensitive information stored by the user in the framework system and to classify the sensitive information into declaration data and content data. The declared data is provided to the AI ​​object, allowing the AI ​​object to read it freely; the content data is maintained through the framework system, preventing the AI ​​object from reading it directly. The AI ​​object reasoning task execution unit is used to execute reasoning tasks based on the declared data and the user's requirements, generate return content, and send the return content to the framework system. The framework system processing unit is used by the framework system to perform set processing operations according to the instructions in the returned content and the set rules, and to obtain the processing results.

5. A sensitive information isolation and processing device for AI objects according to claim 4, characterized in that, In the AI ​​object reasoning task execution unit and the framework system processing unit, the framework system and the AI ​​object interact with each other through a set data protocol; the set data protocol includes: the meaning and format of the declared data received by the AI ​​object; the way the framework system uses sensitive information in the returned content sent by the AI ​​object; and the operations that the framework system needs to perform according to the instructions.

6. The sensitive information isolation and processing device for AI objects according to claim 5, characterized in that, In the sensitive information classification and maintenance unit, the declared data includes: variable name, variable type, and content description; file name, format, size, file content description, and file usage instructions; system version, environment variables, software version, and system setting description information in the runtime environment settings. The content data includes: the variable values; the file itself; and the data files required to build the corresponding runtime environment in the runtime environment settings.

7. A sensitive information isolation and processing device for AI objects according to claim 6, characterized in that, In the AI ​​object reasoning task execution unit, during the process of the AI ​​object performing the reasoning task based on the declared data and the user's requirements, the AI ​​object can read all the declared data. The reading methods include: first, before the framework system sends the Prompt to the AI ​​object, the list of variable declaration data is concatenated into the Prompt; second, the AI ​​object reads the declared data autonomously.

Images