Information processing device, control method for information processing device, program, and storage medium

Abstract

The objective is to provide a mechanism that protects confidential information within text when using AI-generated text, while also obtaining more appropriate results. [Solution] An information processing device comprising: a first conversion means that converts a string to be processed into a converted string in which a protected string contained in the string to be processed is replaced with another string of the same type; an input means that controls the input of the converted string into a generation AI system; an acquisition means that acquires a generated string generated by the generation AI system based on the converted string; a second conversion means that performs a conversion in which a string corresponding to the protected string replaced by the first conversion means is replaced with the protected string before it was replaced by the first conversion means from among the generated strings acquired by the acquisition means; and an output means that outputs the string converted by the second conversion means.

Description

Technical Field

[0001] The present invention relates to a technology for preventing information leakage when using generative AI.

Background Art

[0002] With the recent development of AI technology, generative AI that generates deliverables such as images and texts has been widely used, and printers that provide services using generative AI have also emerged in the field of image forming apparatuses. By the way, some types of generative AI use data input together with generation instructions (prompts) for post-learning (fine-tuning). At this time, if the input data contains confidential information, there is a risk that when a third party uses the generative AI after post-learning, the confidential information of the user may be reflected in the third party's generated product in an unexpected form. In this regard, Patent Document 1 discloses a method for preventing information leakage in a system that performs translation on an AI server, by replacing specific words in a text in a first language with another character string, translating the text into a second language, and replacing the replaced character string in the translated text with a word in the second language.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of Invention

Problems to be Solved by the Invention

[0004] If words corresponding to confidential information in a text are replaced with a character string having no specific meaning such as "AAA" as in the technology of Patent Document 1, it is possible to avoid the above-mentioned risk caused by post-learning. However, in the case of generative AI that understands context after determining attributes such as personal names and company names from words in a text, a meaningless character string such as "AAA" makes it impossible to determine the attributes of the original words, and as a result, appropriate post-learning cannot be performed.

[0005] Therefore, this disclosure aims to provide a mechanism for protecting protected strings within text when using generation AI, and for obtaining more appropriate results. [Means for solving the problem]

[0006] The information processing device according to this disclosure is characterized by comprising: a first conversion means for converting a protected string contained in a string to be processed into a converted string in which a protected string of the same type is replaced; an input means for controlling the input of the converted string into a generation AI system; an acquisition means for acquiring a generated string generated by the generation AI system based on the converted string; a second conversion means for performing a conversion in which a string corresponding to the protected string replaced by the first conversion means is replaced with the protected string before it was replaced by the first conversion means, among the generated strings acquired by the acquisition means; and an output means for outputting the string converted by the second conversion means. [Effects of the Invention]

[0007] According to this disclosure, when using generation AI, it becomes possible to protect protected strings within text and obtain more appropriate results. [Brief explanation of the drawing]

[0008] [Figure 1] A diagram showing an example of a network configuration according to Embodiment 1. [Figure 2] Block diagram showing an example of the hardware configuration of an image forming apparatus. [Figure 3] A flowchart illustrating the general workflow when a user obtains a deliverable using an AI generation service with an image forming apparatus, according to Embodiment 1. [Figure 4] A flowchart illustrating the details of the confidentiality substitution process. [Figure 5] A flowchart showing the details of the confidential information recovery process. [Figure 6] (a) is a diagram showing an example of a replacement table, and (b) is a diagram showing an example of a restoration table. [Figure 7] A flowchart illustrating the general workflow when a user obtains a product using an AI generation service with an image forming apparatus, relating to Modification 1. [Figure 8] (a) to (c) are diagrams showing examples of UI screens. [Figure 9] A flowchart illustrating the general workflow when a user obtains a product using an AI generation service with an image forming apparatus, relating to Modification 2. [Figure 10] A diagram showing an example of a UI screen. [Modes for carrying out the invention]

[0009] The embodiments for implementing this disclosure will be described below with reference to the drawings. The relative arrangement of the components of the apparatus used in the following embodiments, the apparatus shape, etc., are merely examples and are not limiting. Furthermore, in the attached drawings, the same or similar components are given the same reference numeral, and redundant explanations are omitted.

[0010] [Embodiment 1] <System Configuration> Figure 1 shows an example of a network configuration according to this embodiment. In Figure 1, the ONU (Optical Network Unit) 102 is an optical line termination device and is a router-integrated ONU equipped with the function of mutually converting between optical signals and electrical digital signals of the optical line. The router-integrated ONU 102 is connected to the Internet 103 via the optical line.

[0011] The image forming apparatus 100 is a client terminal for using the text generation service provided by the AI ​​generation server 101, and is connected to the internet 103 and the AI ​​generation server 101 via a router-integrated ONU 102. In this embodiment, the image forming apparatus 100 will be described using a multi-function peripheral (Multi-Function Peripheral) equipped with printing and image reading functions as an example. Furthermore, the client terminal does not need to be an image forming apparatus with printing capabilities; it may also be a mobile information processing device such as a smartphone or PDA that can access the AI ​​generation server 101 via a network.

[0012] The generation AI server 101 is a cloud-based server system (generation AI system) that provides a text generation service published on the Internet 103. In this embodiment, it is assumed that the generation AI server 101 generates text, which is the output, based on generation instructions (prompts) from the image forming apparatus 100, which acts as a client terminal, but this is not limited to this. For example, this disclosure can also be applied to an integrated information processing device that combines client and server functions, that is, an information processing device equipped with the functions of the text generation service (generation AI system) provided by the generation AI server 101.

[0013] <Hardware configuration of image forming apparatus> Figure 2 is a block diagram showing an example of the hardware configuration of the image forming apparatus 100 according to this embodiment. The image forming apparatus 100 includes a controller unit 201, a printer unit 202, and a scanner unit 203. The following describes each of the parts shown in Figure 2.

[0014] The controller unit 201 includes a ROM 204, a CPU 205, a RAM 206, a display operation unit 207, and an I / F (LAN) unit 208. The ROM 204 is a non-volatile storage device for storing control execution code (program) of the image forming apparatus 100. The CPU 205 is an arithmetic processing unit that performs overall control of the image forming apparatus 100, and reads out and executes the program stored in the ROM 204. The CPU 205 may be configured with a single processor or a plurality of processors. Further, instead of the CPU 205 or in addition to the CPU 205, hardware such as an ASIC, a DSP, or an FPGA may be provided. In that case, the ASIC or DSP may perform part or all of the processing performed by the CPU 205. The RAM 206 is a work memory for the CPU 205, and is used to expand a program to be executed and temporarily store image data for printing. Although not illustrated, a non-volatile memory (for example, NVRAM) that stores various data required for maintenance of the image forming apparatus 100, information required for print processing, and the like may be provided. The display operation unit 207 is configured with an LCD, an LED, keys, a touch panel, and the like, and receives various operations and instructions from a user via a predetermined user interface screen (UI screen). The I / F (LAN) unit 208 is a network interface that connects to an external device (not illustrated) via a LAN.

[0015] The printer unit 202 performs print processing on a recording sheet using the print image data generated by the CPU 205. The scanner unit 203 optically reads a document to generate image data (scanned image data). Note that the scanner unit 203 includes a document feeder, and can feed a plurality of documents placed on a tray one by one to perform continuous reading. Note that the image forming apparatus 100 may include, for example, a large-capacity storage device represented by an HDD in addition to the components shown in FIG. 2.

[0016] <Operation of Image Forming Apparatus> FIG. 3 is a flowchart illustrating a general operation flow when obtaining a deliverable using a generative AI service in the image forming apparatus 100 according to the present embodiment. The following description will be given along the flowchart of FIG. 3. In the following description, the symbol "S" means a step.

[0017] In S301, the CPU 205 optically reads a document placed on a document platen (not illustrated) using the scanner unit 203 based on a scan instruction from a user, and acquires a scanned image (document image).

[0018] In S302, the CPU 205 performs block selection (character region selection processing) and OCR (character recognition processing) on the document image acquired in S301, and acquires a processing target character string (text) in the document image.

[0019] In S303, the CPU 205 performs confidentiality replacement processing that replaces a protection target character string included in the text acquired in S302, that is, a character string corresponding to confidential information (confidential information character string), with another character string (dummy character string) that allows the type of the character string to be identified. Here, the confidential information includes personal information (name, date of birth, address, telephone number, place of employment, etc.), customer information (name, address, telephone number, card number, etc.), and personnel information (address, family structure, health information, basic salary, bonus amount, etc.). Furthermore, financial information (financial reports, bank transaction statements, etc.), trade secrets (technical know-how, purchase prices, complaint information, etc.), and unpublished product information (product specifications, design information, roadmaps, etc.) may also be included. Details of the confidentiality replacement processing will be described later. Hereinafter, the text in which the confidential information character string has been replaced with the dummy character string is referred to as "replaced text".

[0020] In S304, the CPU 205 transmits the replaced text obtained in S303 to the generative AI server 101 via the I / F (LAN) unit 208. The generative AI server 101 receives the replaced text transmitted from the image forming apparatus 100 as input, executes text processing and generation services such as summarization or translation, and returns a generated character string (generated text) as a deliverable to the image forming apparatus 100.

[0021] In S305, the CPU 205 receives the generated string output from the generation AI server via the I / F (LAN) unit 208.

[0022] In S306, CPU205 performs a desecration process that replaces the dummy strings contained in the generated string received by S305 back into the original confidential information string. Details of this desecration process will be described later.

[0023] In S307, the CPU 205 outputs the generated string (generated text) that has undergone confidentiality decryption processing in S306. Output methods include, for example, display on the display operation unit 207 or printing by the printer unit 202. However, it is not limited to these methods; for example, saving to a BOX or sending a file may also be performed.

[0024] The above outlines the general workflow when obtaining output using the generation AI service in the image forming apparatus 100.

[0025] <Details of confidential information replacement and restoration processes> Next, the details of the confidential information replacement process (S303) and confidential information restoration process (S306) described above will be explained in detail using the flowcharts shown in Figures 4 and 5, with personal information as an example of confidential information. In this embodiment, a conversion table is used in both the replacement with a dummy string in the confidential information replacement process and the restoration to the original confidential information string in the confidential information restoration process. Here, the conversion table used in the confidential information replacement process, which defines the dummy string corresponding to the confidential information to be protected, will be called the "replacement table". Also, the conversion table used in the confidential information restoration process to return the replaced dummy string back to the original confidential information string will be called the "restoration table".

[0026] ≪Details of the confidential information replacement process≫ Figure 4 is a flowchart detailing the confidential information substitution process (S303). The series of processes shown in the flowchart of Figure 4 are realized by the CPU 205 loading a predetermined program stored in ROM 204 into RAM 206 and executing it. The substitution table used in the confidential information substitution process (S303) is prepared in advance on an HDD or the like, and for example, the CPU 205 reads it and loads it into RAM 206 prior to the start of the confidential information substitution process. In this embodiment, it is assumed that the substitution table shown in Figure 6(a), which associates dummy strings with each type (attribute) of personal information, has been read in advance and loaded into RAM 206. The following explanation will follow the flowchart of Figure 4.

[0027] In S401, the CPU 205 refers to the substitution table expanded in RAM 206 and determines whether there are any confidential information strings that have not yet been searched for in the text obtained in S302. In this embodiment, the search for strings corresponding to personal information is performed sequentially for the units of personal information types shown in the substitution table in Figure 6(a), namely "person's name," "date of birth," "address," "telephone number," and "organization (company)." If there are any unprocessed personal information types that have not yet been searched, the process proceeds to S402; otherwise, this process is exited.

[0028] In S402, CPU205 determines which type of data to focus on for processing from among the unprocessed types. In the following S403, CPU205 searches the document obtained in S302 for confidential information strings corresponding to the type of data to focus on determined in S402. For this search, a well-known key-value extraction method is used. Key-value extraction is a method that extracts other strings that are in the vicinity of a key string in a document as value strings corresponding to the key string. For example, if the type of personal information to focus on is "person's name," setting the key string to "name" or "name" will extract strings in the vicinity, such as "Shohei Tanaka," as value strings. Similarly, if the type of personal information to focus on is "date of birth," setting the key string to "date of birth" or "birthday" will extract strings in the vicinity, such as "June 19, 2008," as value strings. Depending on the type of data to focus on, conditions such as the number of characters and the format of the string to be extracted may also be added. In this way, a process is performed to search the text for personal information strings that correspond to the type of information of interest.

[0029] In S404, CPU205 determines whether it was able to find a string of confidential information of the type of interest as a result of the search in S403. If a string of confidential information of the type of interest is found within the document, the process proceeds to S405; otherwise, it returns to S401.

[0030] In S405, CPU205 refers to the substitution table and replaces confidential information strings of the attention type found in the document with dummy strings associated with that attention type. For example, suppose "Tanaka Shohei" is found as personal information string corresponding to the attention type "Person's Name". In this case, in the substitution table in Figure 6(a), it will be replaced with the dummy string "Yamada Taro", which is associated with the type "Person's Name" and has substitution ID=1. As shown in Figure 6(a), if confidential information is personal information, one substitution ID is assigned to the personal information of one person. Therefore, if the personal information string "Tanaka Shohei" is found two or more times in the document, all of those strings will be replaced with the dummy string "Yamada Taro". This is because "Tanaka Shohei" is a personal information string that represents the name of the same specific individual. In response to this, if another personal information string, "Kawamura Takuya," corresponding to the "personal name" category is also found, this string will be replaced with a dummy string associated with the "personal name" category and with a different replacement ID (for example, the string "Sato Jiro" with replacement ID=2). After performing this conversion to a dummy string, CPU205 generates a recovery table so that the converted string can be restored to the original string in the subsequent confidential information restoration process. Specifically, it stores information indicating what kind of conversion was performed (i.e., the personal information string before replacement, the dummy string after replacement, the category, and the replacement ID) in the recovery table. In this way, a recovery table like the one shown in Figure 6(b) is obtained. After the replacement to a dummy string and the generation of the recovery table that accompanies it, the process returns to S401.

[0031] The above describes the confidential information replacement process. In this embodiment, personal information strings are identified by key-value extraction, and these identified personal information strings are replaced with dummy strings using a replacement table, but the embodiment is not limited to this. For example, a trained model obtained by training on a large amount of text in which personal information strings contained in the text have been replaced with dummy strings may be used to perform the identification of personal information strings and their replacement with dummy strings in one step. Furthermore, the above confidential information replacement process may be performed by an external device (not shown) based on a request from the image forming apparatus 100, and the image forming apparatus 100 may receive the results from the external device. In addition, the replacement table may be editable by the user.

[0032] ≪Details of the confidential information restoration process≫ Figure 5 is a flowchart detailing the confidential data decryption process (S306). The series of processes shown in the flowchart of Figure 5 are realized by the CPU 205 loading a predetermined program stored in the ROM 204 into the RAM 206 and executing it. As mentioned above, the decryption table used in the confidential data decryption process (S306) is assumed to have been generated in the preceding confidential data replacement process and loaded into the RAM 206. In this embodiment, it is assumed that the decryption table shown in Figure 6(b) is loaded into the RAM 206. The following explanation will follow the flowchart of Figure 5.

[0033] In S501, the CPU 205 refers to the restoration table expanded in RAM 206 and determines whether there are any dummy strings that have not yet been searched within the document, which is the output obtained in S305. In this embodiment, the dummy string search is performed sequentially for each replacement ID shown in the restoration table in Figure 6(b). If there are replacement IDs that have not yet been searched, the process proceeds to S502; otherwise, this process is exited.

[0034] In S502, CPU205 determines which replacement IDs to focus on for processing from among the unprocessed replacement IDs. In the next step, S503, CPU205 searches for a dummy string related to the replacement ID of interest determined in S502 within the text, which is the output obtained in S305.

[0035] In S504, CPU205 determines whether it was able to find a dummy string related to the target replacement ID as a result of the search in S503. If a dummy string related to the target replacement ID is found in the text, the process proceeds to S505; otherwise, it returns to S501.

[0036] In S505, CPU205 refers to the recovery table and replaces the dummy strings related to the attention replacement ID found in the document with the original protected strings, i.e., the corresponding personal information strings. For example, in the recovery table in Figure 6(b), when attention replacement ID = 1, the dummy string "Yamada Taro" of type "Person's Name" is converted to the original personal information string "Tanaka Shohei". Furthermore, the dummy string "Kakuu Co., Ltd." of type "Organization (Company)" is converted to the original personal information string "Shozoku Co., Ltd.". This is the content of the confidential information recovery process.

[0037] <Example 1> The user may be asked to confirm the replaced text containing the dummy string obtained by the confidential replacement process (S303) before sending it to the generation AI server. Figure 7 is a flowchart showing the general flow of operations when obtaining deliverables using the generation AI service according to this modified example. The only difference from the flowchart in Figure 3 according to the above embodiment is the addition of S701 and S702, so these steps, which represent the differences, will be explained below.

[0038] In S701, the CPU 205 displays the result of the confidential replacement process obtained in S303 on the display operation unit 207. Figure 8(a) shows an example of a UI screen including a replacement document. On such a UI screen, the user checks whether the dummy string in the replacement document is appropriate, and presses the OK button if there is no problem, or the Cancel button if there is a problem.

[0039] In S702, the CPU 205 determines the next process to execute based on the user's button input. If the CPU 205 detects a user authorization operation (pressing the OK button), it proceeds to S304 and sends the replacement text to the generation AI server 101. At this time, the CPU 205 displays a UI screen (see Figure 8(b)) on the display operation unit 207 that includes a message indicating that the replacement text has been sent to the generation AI server 101. Through this UI screen, the CPU 205 informs the user that the replacement text has been sent to the generation AI server 101 and prompts them to wait until the output is sent from the generation AI server 101. On the other hand, if the CPU 205 detects a user denial operation (pressing the Cancel button), it exits this process without sending the replacement text to the AI ​​server 101. At this time, the CPU 205 displays a UI screen (see Figure 8(c)) on the display operation unit 207 that includes a message indicating that the use of the generation AI service has been canceled. Through this UI screen, the CPU 205 informs the user that the use of the generation AI service has been canceled. The user can then return to the home screen by pressing the home button on the UI screen shown in Figure 8(c). Whether the OK or Cancel button has been pressed can be detected by the user's touch operation on the touch panel, or by the operation signal from an input device such as a mouse or keyboard. This concludes the description of this modified example.

[0040] <Modification 2> In the above-described modification 1, the user was allowed to confirm the result of the confidentiality replacement process in advance, and if it was deemed unacceptable, the process was terminated without sending the replaced document to the AI ​​server 101. However, it may be possible to allow the user to modify the dummy string in the replaced document. Figure 9 is a flowchart showing the general flow of operations when obtaining deliverables using the generation AI service in this modification. The difference from the flowchart in Figure 7 related to modification 1 is that S901 has been added. The differences will be explained below.

[0041] In this modified example, if an unauthorized user operation (pressing the cancel button) is detected in S702, the CPU 205 transitions the user from the UI screen shown in Figure 8(a) on the display operation unit 207 to a UI screen where the user can manually correct the problematic part of the replacement text. Figure 10 shows an example of the UI screen after the transition, which includes a software keyboard for the user to manually correct the problematic part of the replacement text. Through this UI screen, the user checks the displayed replacement text and corrects the dummy string using the software keyboard. At this time, the user may also correct strings other than the dummy string. When the CPU 205 receives the user's correction operation of the dummy string, etc., it reflects the content of the correction operation in the replacement text and sends the corrected replacement text to the generation AI server 101 in S304. This concludes the description of this modified example.

[0042] <Other variations> Although the present invention has been described in detail based on its preferred embodiments, the present invention is not limited to these specific embodiments, and various forms that do not depart from the spirit of the invention are also included in the present invention. Furthermore, each of the embodiments described above is merely one embodiment of the present invention, and it is possible to combine each embodiment as appropriate. In addition, although the embodiments described above have been explained using the application of the present invention to an MFP as an example, the present invention is not limited thereto. The present invention is broadly applicable to electronic devices having an imaging means (camera function) capable of digitizing a document. Examples of such electronic devices include notebook PCs, PDAs, smartphones, portable image viewers, digital photo frames, music players, game consoles, e-book readers, and network cameras. Furthermore, even if the present invention does not have an imaging means itself, it is applicable to devices that can communicate with the above-mentioned electronic devices having an imaging means and acquire document images from said devices via wired or wireless means, such as printer devices with displays.

[0043] (Other embodiments) The present invention can also be realized by performing the following process: supplying software (programs) that realize the functions of the embodiments described above to a system or device via a network or various storage media, and having the computer (or CPU, MPU, etc.) of that system or device read and execute the program code. In this case, the program and the storage medium storing the program constitute the present invention.

[0044] Furthermore, this disclosure includes the following configurations and methods.

[0045] [Configuration 1] A first conversion means that converts the string to be processed into a converted string in which the protected string contained in the string to be processed is replaced with another string of the same type, An input means for controlling the input of the converted string into the generation AI system, An acquisition means for acquiring a generated string that is generated by the generation AI system based on the converted string, A second conversion means performs a conversion in which, among the generated strings obtained by the acquisition means, the string corresponding to the protected string that was replaced by the first conversion means is replaced with the protected string before it was replaced by the first conversion means. Output means for outputting the string converted by the second conversion means, An information processing device characterized by having the following features.

[0046] [Configuration 2] The information processing apparatus according to configuration 1, characterized in that the protected string is a string representing personal information.

[0047] [Configuration 3] The information processing apparatus according to configuration 1 or 2, characterized in that the first conversion means converts the protected string to the converted string using a first table which associates the converted string with each type of personal information.

[0048] [Structure 4] The first table above associates multiple converted strings with different content for each type of personal information. The first conversion means described above is: If the specific individual corresponding to the protected string is the first individual, then the string is converted to the first of the multiple converted strings. If the specific individual corresponding to the protected string is a second individual, then convert to the second converted string from among the multiple converted strings. The information processing device according to configuration 3, characterized by the above.

[0049] [Composition 5] The information processing apparatus according to configuration 3, characterized in that the second conversion means performs a conversion in which it replaces the string corresponding to the string to be protected from the generated string with the string to be protected from the original string, using a second table which associates the converted string after replacement by the conversion in the first conversion means with the string to be protected from the original string.

[0050] [Composition 6] The information processing apparatus according to any one of configurations 1 to 5, characterized in that the string to be processed is a string extracted from an image obtained by capturing or scanning a document.

[0051] [Composition 7] The information processing apparatus according to configuration 6, characterized in that the information processing apparatus has an imaging means.

[0052] [Structure 8] The information processing apparatus according to configuration 6, characterized in that the information processing apparatus has a scanning means.

[0053] [Composition 9] The information processing apparatus according to configuration 8, characterized in that the information processing apparatus is further a printer having printing means.

[0054] [Configuration 10] The information processing apparatus according to any one of configurations 1 to 9, characterized in that the output means performs display or printing as output.

[0055] [Composition 11] The information processing apparatus according to any one of configurations 1 to 9, characterized in that the input means displays the converted string, controls the generation AI system to input the converted string if permission is granted by the user, and controls the generation AI system not to input the converted string if permission is not granted.

[0056] [Composition 12] The information processing apparatus according to any one of configurations 1 to 9, characterized in that the input means displays the converted string, accepts a user's operation to modify the string, and then controls the system to input the modified converted string into the generation AI system.

[0057] [Method 1] A first conversion step involves converting the string to be processed into a converted string in which the protected string contained in the string to be processed is replaced with another string of the same type, An input step that controls the system to input the converted string into the generation AI system, A step of obtaining a generated string that is generated by the generation AI system based on the converted string, A second conversion step is performed to replace the strings corresponding to the protected strings that were replaced in the first conversion step with the protected strings that were replaced in the first conversion step, among the generated strings obtained in the acquisition step, An output step that outputs the string converted in the second conversion step described above. A control method for an information processing device, characterized by having the following features.

[0058] [Composition 13] A program for causing a computer to function as one of the information processing devices described in any one of the configurations 1 to 12.

[0059] [Composition 14] A computer-readable storage medium containing a program for causing the computer to function as one of the information processing devices described in any one of items 1 to 12.

Claims

1. A first conversion means that converts the string to be processed into a converted string in which the protected string contained in the string to be processed is replaced with another string of the same type, An input means for controlling the input of the converted string into the generation AI system, An acquisition means for acquiring a generated string generated by the generation AI system based on the converted string, A second conversion means performs a conversion in which, among the generated strings obtained by the acquisition means, the string corresponding to the protected string that was replaced by the first conversion means is replaced with the protected string before it was replaced by the first conversion means. Output means for outputting the string converted by the second conversion means, An information processing device characterized by having the following features.

2. The information processing apparatus according to claim 1, characterized in that the protected string is a string representing personal information.

3. The information processing apparatus according to claim 2, characterized in that the first conversion means converts the protected string to the converted string using a first table which associates the converted string with each type of personal information.

4. The first table above associates multiple converted strings with different content for each type of personal information. The first conversion means is, If the specific individual corresponding to the protected string is the first individual, then the string is converted to the first of the multiple converted strings. If the specific individual corresponding to the protected string is a second individual, then convert to the second converted string from among the multiple converted strings. The information processing apparatus according to claim 3.

5. The information processing apparatus according to claim 3, characterized in that the second conversion means performs a conversion in which it replaces the string corresponding to the string to be protected from the generated string with the string to be protected from the protected string from the generated string, using a second table that associates the converted string after replacement by the conversion in the first conversion means with the string to be protected from the protected string from the protected string from the protected string from the first replacement.

6. The information processing apparatus according to claim 1, characterized in that the string to be processed is a string extracted from an image obtained by capturing or scanning a document.

7. The information processing apparatus according to claim 6, characterized in that the information processing apparatus has an imaging means.

8. The information processing apparatus according to claim 6, characterized in that the information processing apparatus has a scanning means.

9. The information processing apparatus according to claim 8, characterized in that the information processing apparatus is further a printer having printing means.

10. The information processing apparatus according to claim 1, characterized in that the output means performs display or printing as output.

11. The information processing apparatus according to claim 1, characterized in that the input means displays the converted string, controls the generation AI system to input the converted string if permission is granted by the user, and controls the generation AI system not to input the converted string if permission is not granted.

12. The information processing apparatus according to claim 1, characterized in that the input means displays the converted string, accepts a user's operation to modify the string, and then controls the system to input the modified converted string into the generation AI system.

13. A first conversion step involves converting the string to be processed into a converted string in which the protected string contained in the string to be processed is replaced with another string of the same type, An input step that controls the system to input the converted string into the generation AI system, An acquisition step of obtaining a generated string generated by the generation AI system based on the converted string, A second conversion step is performed to replace the string corresponding to the protected string that was replaced in the first conversion step with the protected string before it was replaced in the first conversion step, among the generated strings obtained in the acquisition step. An output step which outputs the string converted in the second conversion step described above. A control method for an information processing device, characterized by having the following features.

14. A program for causing a computer to function as one of the means of an information processing apparatus described in any one of claims 1 to 12.

15. A computer-readable storage medium storing a program for causing a computer to function as one of the means of an information processing device described in any one of claims 1 to 12.

Images