Generative artificial intelligence content detection method, related device and storage medium

Abstract

The present application relates to the field of artificial intelligence, and provides a generative artificial intelligence content detection method, a related device and a storage medium, the method comprising: obtaining a first input of a first user on a detection interface, the first input indicating a to-be-detected content, the to-be-detected content comprising data in at least one state of a text, an image, a video or a webpage, determining a target detection mode according to the media type corresponding to the to-be-detected content, and performing forgery detection on the to-be-detected content; and if it is detected that the target element meeting the preset counterfeiting condition exists, generating an authentication result and outputting an instruction. According to the scheme, multi-dimensional authentic identification can be carried out on the to-be-detected content in all directions in various mainstream terminal user-oriented scenes, so that the authentic identification result is more comprehensive and more accurate, and the method is suitable for more scenes. According to the method, the to-be-detected content of any module combination can be authenticated by adopting at least one test mode, so that a terminal user can conveniently and timely perform real-time counterfeiting detection on the media content in different modes on a terminal side, and the terminal user is prevented from being cheated.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of generative artificial intelligence content detection methods, and in particular, to a generative artificial intelligence content detection method, related devices and storage media. Background Art

[0002] At present, various methods and software of generative artificial intelligence have been widely used, which has brought about potential risks and hidden dangers. In addition to the leakage of personal information, negative news, false news and speeches of key figures produced by generative artificial intelligence tools are difficult to distinguish between true and false, spread rapidly, and seriously mislead public opinion, affecting national security and social stability. It may also become a new means of cybercrime and infringe on the legitimate rights and interests of citizens.

[0003] In the process of research and practice of the prior art, the inventor of the embodiment of the present application found that currently only images and videos generated based on deep synthesis technology can be intelligently detected on the market. With the abuse of AIGC, especially the increasing number of forgeries in text and audio, images generated based on diffusion models and texts generated based on large language models LLM cannot be detected by general deep forgery detection technology, and cannot be distinguished manually, which brings great obstacles to the identification of judicial evidence. In addition, current products can only detect forgeries of offline media content, and are limited to the server side. Now users are more inclined to read various media content on the terminal, and have voice or video conversations, but can only analyze the forgery behavior afterwards. Usually, voice or video conversations based on social software such as interactive applications will not have recorded offline internal content, so this makes it more difficult to extract evidence, and also leads to the inability to detect in real time that one party in the conversation is forged. Summary of the invention

[0004] The embodiments of the present application provide a generative artificial intelligence content detection method, related devices and storage media, which can improve the efficiency of the generative artificial intelligence content detection method, and can perform multi-dimensional authentication of the content to be detected in a variety of mainstream application scenarios for terminal users, so that the authentication results are more comprehensive, more accurate, and applicable to more scenarios. Regardless of whether the content to be detected contains one type of media content or includes media content of more than two modes, at least one test method can be used to authenticate any combination of the content to be detected, which is convenient for terminal users to perform real-time forgery detection of media content in different modes on the terminal side in a timely manner to avoid being deceived.

[0005] In a first aspect, the present application provides a generative artificial intelligence content detection method from the perspective of a generative artificial intelligence content detection method, the method comprising:

[0006] Acquire a first input of a first user in the detection interface, where the first input indicates content to be detected, and the content to be detected includes data in at least one state of text, image, video, or web page;

[0007] In response to the first input, determining a target detection mode according to a media type corresponding to the content to be detected, and performing forgery detection on the content to be detected under the target detection mode;

[0008] If it is detected that there is a target element that meets the preset forgery condition in the content to be detected, the authentication result is output and the first prompt information indicating the authentication result is output, and the authentication result indicates that at least one target element is forged content, and the preset forgery condition is data generated by at least one of generating adversarial networks, generating based on diffusion models, or generating based on large language models.

[0009] In some implementations, if a second user and a non-second user are present in the real-time target session and both generate corresponding media content in the real-time target session, then after detecting the presence of a voice that does not belong to the real second user in the real-time target session, the duration and appearance time of the media content corresponding to the second user and the non-second user are analyzed and marked, and a second prompt message is output.

[0010] In a second aspect, an embodiment of the present application provides a generative artificial intelligence content detection device, which has the function of implementing the generative artificial intelligence content detection method provided in the first aspect above. The function can be implemented by hardware, or by hardware executing the corresponding software. The hardware or software includes one or more modules corresponding to the above functions, and the modules can be software and / or hardware.

[0011] In some implementations, the generative artificial intelligence content detection device includes:

[0012] An input-output module, configured to obtain a first input from a first user on the detection interface, wherein the first input indicates content to be detected, and the content to be detected includes data in at least one state of text, image, video or web page;

[0013] A processing module, configured to determine, in response to the first input, a target detection mode according to a media type corresponding to the content to be detected, and perform forgery detection on the content to be detected under the target detection mode;

[0014] The processing module is also used to generate a forgery detection result and output a first prompt information indicating the forgery detection result through the input-output module if a target element that meets a preset forgery condition is detected in the content to be detected, wherein the forgery detection result indicates that at least one target element is forged content, and the preset forgery condition is data generated by at least one of generating an adversarial network, generating an anti-difference model, or generating an anti-difference based on a large language model.

[0015] In a fourth aspect, an embodiment of the present application provides a generative artificial intelligence content detection device, which includes: at least one processor and a memory; wherein the memory is used to store a computer program, and the processor is used to call the computer program stored in the memory to execute the steps in the above-mentioned first aspect, any one of the implementation methods of the first aspect, or any one of the generative artificial intelligence content detection methods provided in the second aspect.

[0016] In a fifth aspect, an embodiment of the present application provides a computer-readable storage medium having a function of implementing the generative artificial intelligence content detection method provided in the first aspect above. The function can be implemented by hardware, or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions, and the modules can be software and / or hardware. Specifically, the computer-readable storage medium stores a plurality of instructions, and the instructions are suitable for loading by a processor to execute the steps in the first aspect, any implementation of the first aspect, or any generative artificial intelligence content detection method provided in the second aspect in the embodiment of the present application.

[0017] Compared with the prior art, the solution provided in the embodiment of the present application can realize offline forgery detection and real-time online forgery detection of media content in diversified scenarios by adopting the method provided in the embodiment of the present application. It can dynamically change based on different application scenarios generated by media content and is suitable for multimodal media content detection. For users, both offline forgery detection and online real-time forgery detection can be realized. The combination of the two methods can provide detection information to users in a timely manner, remind users, and reduce users from being deceived or guided by forged generative artificial intelligence content. At the same time, it also meets the user's demand for forgery detection of the massive amount of media content received every day, without relying on third-party detection platforms or authoritative detection agencies, and the user experience is extremely high. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic diagram of a flow chart of a generative artificial intelligence content detection method in an embodiment of the present application;

[0019] Figure 2aA schematic diagram of an interface for detecting forgery of media content in a voice conversation in an embodiment of the present application;

[0020] Figure 2b A schematic diagram of an interface for detecting forgery of media content in a video session in an embodiment of the present application;

[0021] Figure 3a This is a schematic diagram of an interface for detecting forgery of text content in an embodiment of the present application;

[0022] Figure 3b A schematic diagram of an interface for detecting forgery of image content in an embodiment of the present application;

[0023] Figure 3c A schematic diagram of an interface for detecting forgery of video content in an embodiment of the present application;

[0024] Figure 4 A schematic diagram of an interface display for dynamic forgery detection of page turning in a web page in an embodiment of the present application;

[0025] Figure 5 This is a schematic diagram of the structure of a generative artificial intelligence content detection device in an embodiment of the present application;

[0026] Figure 6 It is a structural schematic diagram of a physical device for implementing a generative artificial intelligence content detection method in an embodiment of the present application;

[0027] Figure 7 It is a structural schematic diagram of a mobile phone that implements the generative artificial intelligence content detection method in the embodiment of the present application. DETAILED DESCRIPTION

[0028] The terms "first", "second", etc. in the specification and claims of the embodiments of the present application and the above-mentioned drawings are used to distinguish similar objects (for example, the first user and the second user in the embodiments of the present application respectively represent different users participating in the conversation), and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described here can be implemented in an order other than the content illustrated or described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or modules is not necessarily limited to those steps or modules clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices. The division of modules that appear in the embodiments of the present application is only a logical division. There may be other division methods when implemented in actual applications. For example, multiple modules can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between each other shown or discussed can be through some interfaces, and the indirect coupling or communication connection between modules can be electrical or other similar forms, which are not limited in the embodiments of the present application. Furthermore, the modules or sub-modules described as separate components may or may not be physically separated, may or may not be physical modules, or may be distributed in multiple circuit modules, and some or all of the modules may be selected according to actual needs to achieve the purpose of the embodiment of the present application.

[0029] The embodiments of the present application provide a generative artificial intelligence content detection method, related devices and storage media, which can be used in a server or terminal device. Specifically, it can be used to detect the authenticity of mainstream generative artificial intelligence content, so as to distinguish it from artificially written texts, real images and videos taken by a camera, and real sounds spoken by real people. It can also detect the forgery of media content (such as sound and human face images) generated in real-time voice conversations and video conversations.

[0030] In some implementations, the present solution is applied to a generative artificial intelligence content detection device, which includes the following functional modules:

[0031] 1) Detection content input module: This module supports the input of video, image, audio and text file types. When uploading, users can upload data by checking the options of image, video, audio or text. At the same time, it also supports the import of single files or batch files. For imported files, the product will check whether the files meet the requirements in terms of file size, audio and video length, etc. For files that do not meet the requirements, the user will be prompted with the reason for the error to facilitate the user to reprocess and upload again. For files that meet the requirements, you can enter the subsequent process.

[0032] 2) Algorithm detection module: The algorithm detection stage is to perform synthetic detection on the files to be detected input by the user in the queue order. After the algorithm reasoning operation is completed, the detection results will be output for the user to view.

[0033] 3) Result display module: After the algorithm has completed the detection of all the files imported by the user, it will output the detection results. On the detection result page, the detection results of all images, audio or text will be displayed. Users can view the authenticity judgment results, images, video frames, audio, text detection generated images and other information here. At the same time, users are supported to download the detection report online.

[0034] The generative AI content detection device can detect forgery of videos, images or audios disseminated by the public in any form (e.g., on social platforms, social software, or on web pages), and can also detect forgery of media content generated in communication sessions (e.g., voice calls, video calls, phone card-based calls, etc.) to identify the identity of the other party in the communication session. Application scenarios include: detecting the authenticity of audio and video evidence, detecting the compliance of network content, combating infringement of personal reputation, combating network fraud, and improving the security of biometric systems.

[0035] The solutions provided in the embodiments of the present application involve technologies such as artificial intelligence (AI), natural language processing (NLP), and machine learning (ML), which are specifically described by the following embodiments:

[0036] Among them, AI is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines so that machines have the functions of perception, reasoning and decision-making.

[0037] AI technology is a comprehensive discipline that covers a wide range of fields, including both hardware-level and software-level technologies. The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics and other technologies. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0038] NLP is an important direction in the fields of computer science and artificial intelligence. It studies various theories and methods that can achieve effective communication between people and computers using natural language. Natural language processing is a science that integrates linguistics, computer science, and mathematics. Therefore, research in this field will involve natural language, that is, the language people use in daily life, so it is closely related to the study of linguistics. Natural language processing technology usually includes text processing, semantic understanding, machine translation, robot question answering, knowledge graph and other technologies.

[0039] Among them, it should be particularly noted that the generative artificial intelligence content detection device involved in the embodiments of the present application or the terminal installed with the generative artificial intelligence content detection device can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, a personal digital assistant, a vehicle terminal, a wearable device (such as smart glasses, smart bracelets, VR helmets), etc., but is not limited to this. The generative artificial intelligence content detection device and the server can be directly or indirectly connected by wired or wireless communication, and the embodiments of the present application are not limited here.

[0040] When users are on various social platforms or using various communication software, they will encounter abuse based on AIGC, especially forged media content in text and audio, and even impersonating communication friends to communicate and interact with themselves. These media contents generated by AIGC cannot be detected by general deep fake detection technology, and cannot be distinguished manually, which brings great obstacles to the identification of judicial evidence. In addition, current products can only detect forgery of offline media content, and are limited to the server side. Now users prefer to read various media content on the terminal, and have voice or video conversations, but can only analyze the forgery behavior afterwards. Usually, voice or video conversations based on social software such as interactive applications will not have recorded offline internal content, so this makes it more difficult to extract evidence, and also leads to the inability to detect in real time that one party to the conversation is forged. In this regard, the embodiments of the present application mainly adopt the following technical solutions:

[0041] 1. Provide a multi-modal forgery detection method for end users, which can be applied in offline or online scenarios, and can also be used for identity recognition in voice or video conversations.

[0042] 2. Provide users with multiple interaction modes.

[0043] The following combination Figure 1-Figure 7 The technical solution of the embodiment of the present application is exemplified.

[0044] Figure 1 This is a flow chart of a generative artificial intelligence content detection method in an embodiment of the present application. The execution subject of the method is a generative artificial intelligence content detection device, or a first terminal installed with a generative artificial intelligence content detection device, and the present application embodiment is not limited to this. The present application embodiment includes:

[0045] 101. Obtain a first input of a first user on the detection interface.

[0046] The first input indicates content to be detected, and the content to be detected includes data in at least one state of text, image, video or web page.

[0047] It should be noted that in the embodiments of the present application, the content to be detected may include data generated by any forgery method, or any combination of forgery methods. The embodiments of the present application only illustrate the current mainstream data generated based on generative adversarial networks, data generated based on diffusion models, and data generated based on large language models. The embodiments of the present application are not limited to this. The generative artificial intelligence content detection method can arbitrarily expand the functional module of "preset detection mode for forgery detection of data with new forgery methods".

[0048] 102. In response to the first input, determine a target detection mode according to a media type corresponding to the content to be detected, and perform forgery detection on the content to be detected in the target detection mode.

[0049] 103. If it is detected that a target element that meets a preset forgery condition exists in the content to be detected, a forgery detection result is generated and first prompt information indicating the forgery detection result is output.

[0050] The authentication result indicates that at least one target element is forged content, and the preset forgery condition is data generated by at least one of a generative adversarial network, a diffusion model, or a large language model.

[0051] The authentication result can show the detection result corresponding to the data under at least one preset forgery method in the data to be tested, and each detection result includes whether the content to be tested meets the preset forgery conditions, score, confidence, forgery method, content traceability, etc. For example, the detection result can give corresponding detection results according to different detection tasks (such as Deepfake, Diffusion, and LLM detection modes), including: judgment results, detection scores, etc. Among them, in terms of forged content detection, Deepfake, Diffusion, and LLM synthetic data can be detected, among which Deepfake synthetic data supports the detection of synthetic faces in videos and images and the detection of synthetic voices in audio, and Diffusion detection supports the detection of synthetic faces in images.

[0052] In some implementations, the first prompt information is output in at least one of the following ways:

[0053] Display in the form of a prompt window;

[0054] Displayed as a callout box;

[0055] Display with preset annotation shapes;

[0056] Display as a web message;

[0057] Or output in voice broadcast mode.

[0058] It can be seen that by providing a variety of forged result prompting methods, the user can be prompted in a targeted manner to where the forged content is located, and the forged content can be located intuitively and quickly.

[0059] Optionally, in some embodiments of the present application, the first terminal currently runs an interactive application, and the first user is in a real-time target conversation in a target mode with a second user of at least one second terminal, and the target mode includes a voice conversation or a video conversation; the method further includes:

[0060] When it is detected that the first terminal starts the interactive application and the first user and the second user are in a real-time target session in a target mode;

[0061] The first terminal can perform forgery detection on any media content generated in the real-time target session in the current target mode, either through a preset voice detection mode or based on the real-time voiceprint feature of the second user and the preset voice detection mode. Specifically, forgery detection can be performed in the following two situations: voice session and video session:

[0062] Case 1: performing forgery detection on the real-time voice of at least one of the second users in the voice conversation.

[0063] Specifically, first historical data of the second user is obtained, where the first historical data is obtained based on voice data of the second user in a voice conversation between the first user and the second user within a historical period; and based on the first historical data and a preset voice detection mode, forgery detection is performed on at least one real-time voice of the second user in the voice conversation.

[0064] It can be understood that the voice conversation mentioned in the embodiments of the present application may include voice chat based on interactive applications, voice conferences based on web pages, applets or communication applications, and may also include voice calls or video calls based on telephone communications using phone cards, and the specific embodiments of the present application are not limited to this.

[0065] For example, voice data of a voice conversation between the first user and the second user within one year can be obtained in advance, and then the voice data can be input into a voice-related model for analysis to obtain and store the real voiceprint features that can characterize the second user. When the first user starts a voice conversation with the second user, the first terminal can directly extract the voiceprint features of the second user's current voice conversation, and then compare the extracted voiceprint features with the real voiceprint features. If the similarity is lower than the first similarity, it is determined that the second user in the current voice conversation is a user with a fake identity, not the real second user.

[0066] For example, Figure 2a The first user's voice conversation interface, where user a is the first user, and user b, user c, and user d are all second users participating in the voice conversation. User d is currently speaking, and the generative artificial intelligence content detection device detects whether the voiceprint features of user d are forged. Through analysis, it is concluded that the current user d speaking is not actually the real user d's voice, so the first user is prompted, as well as the other second users participating in this voice conversation (i.e., user b and user c) except user d: the current user d speaking may not be the real user, with a probability of 0.92.

[0067] Case 2: Alternatively, forgery detection is performed on at least one video character image of the second user in the video session.

[0068] Specifically, second historical data of the second user is obtained, where the second historical data is obtained based on video data of the second user in a video conversation between the first user and the second user within a historical period; and based on the first historical data and a preset video detection mode, forgery detection is performed on at least one real-time video image of the second user in the video conversation.

[0069] It can be understood that the video conversation mentioned in the embodiments of the present application may include video chat based on interactive applications, video conferencing based on web pages, applets or communication applications, and the embodiments of the present application are not limited to this.

[0070] For example, the video data of the video conversation between the first user and the second user within one year can be obtained in advance, and then the voice data can be input into the voice-related model for analysis to obtain and store the real voiceprint features that can characterize the second user; or, the second user's head portrait, or the photos of the second user or the group photos containing the second user's face that appear in the daily single chat between the first user and the second user or the group chat in which the second user participates can be obtained in advance, and then the T-zone features of the second user's face are analyzed, and the real face features characterizing the second user obtained by the analysis are stored. When the first user starts a video conversation with the second user, the first terminal can directly extract the face features of the second user in the current video conversation, and then compare the extracted real-time face features with the real face features. If the similarity is lower than the second similarity, it is determined that the second user in the current video conversation is a user with a fake identity and is not the real second user.

[0071] For example, Figure 2b The video conversation interface of the first user, wherein user a is the first user, and user b, user c and user d are all second users participating in the video conversation. User d is currently speaking, and the generative artificial intelligence content detection device detects whether at least one of the facial features and voiceprint features of user d is forged. Through analysis, it is concluded that the user d currently speaking is not actually the face or voice of the real user d, so the first user is prompted, as well as the second users (i.e., user b and user c) participating in this video conversation except user d: the user d currently speaking may not be the real person, with a probability of 0.92.

[0072] It is understandable that the solution can also be used in a monitoring APP to enable real-time monitoring video conversations in intercom mode. The first terminal is the monitoring end, which watches the real-time monitoring screen transmitted by the camera, and performs forgery detection on the first terminal for the characters appearing in the real-time monitoring screen.

[0073] In the above case 1 and case 2, if it is detected that at least one of the second users in the voice conversation meets the preset forgery condition, or if it is detected that at least one of the second users in the video conversation meets the preset forgery condition, a first mark is set in the voice conversation interface or the video conversation interface, and the first mark is used to indicate the confidence that the second user is not the person himself. In case 2, the first historical data in case 1 can be combined with the second historical data to determine whether the second user is the person himself.

[0074] If the voice or video image of the second user in the current real-time target conversation is generated based on deep synthesis technology or generative artificial intelligence related technology, then the media content (such as voice, video image) generated by the second user will definitely contain traces of forgery, and this solution will definitely be able to detect it.

[0075] It can be seen that, on the one hand, through this solution, it is possible to authenticate the identity of any party (the other party is taken as an example in the embodiment of the present application) participating in the communication interaction between terminals (such as voice conversations and video conversations) to ensure that the party generating the media content in the communication interaction is truly from the second user. On the other hand, the above-mentioned situation 1 and situation 2 can be combined, and both can be implemented separately. In actual detection, the media content transmitted to the second user by the interactive application currently used by any party (such as the second user) is detected for forgery. It is not limited to the situation where the second user side is a non-second user in the entire real-time target session, or the second user and the non-second user are present and both generate media content in the real-time target session (for example, there may be a scenario where the second user is coerced by the non-second user). The embodiments of the present application are all applicable.

[0076] If in the real-time target session, there is a second user and a non-second user who both generate media content in the real-time target session, then even if the second user and the non-second user generate media content separately, after detecting that the real-time voiceprint features of a certain time period do not belong to the real second user, the duration, appearance time, etc. of the media content of the second user and the non-second user can be analyzed and marked, so as to facilitate the subsequent analysis of evidence.

[0077] Optionally, in some embodiments of the present application, the solution can also be used to detect forgery of any modal media content appearing in a web page when a user browses the web page. Specifically, the first terminal displays a real-time web page browsed by the first user, and the method further includes:

[0078] Following the changes in the real-time webpage content displayed when the first user browses the real-time webpage, the target area in the real-time webpage content that meets the preset counterfeiting condition is detected in real time; and a corresponding second mark is dynamically set for at least one target area detected in real time.

[0079] The second mark includes at least one of the following items:

[0080] 1. A first confidence level indicating that the text in the target area meets the preset forgery condition, for example, indicating that there are abnormal paragraphs in the text in the target area that meet the preset forgery condition. Figure 3a shown.

[0081] 2. A second confidence level indicating that the image in the target area meets the preset forgery condition, for example, indicating that there is an abnormal image in the target area that meets the preset forgery condition. Figure 3b shown.

[0082] 3. A third confidence level indicating that the video played in the target area meets the preset forgery condition, for example, indicating that there are abnormal video segments that meet the preset forgery condition in the video played in the target area. Figure 3c shown.

[0083] It is understandable that the generative artificial intelligence content detection device can track the browsing behavior of the first user, follow the changes in the web page content corresponding to the browsing trajectory of the first user (that is, the media content of any modality that appears), and start the preset detection mode corresponding to the media content of the corresponding modality in the web page content to perform forgery detection on the media content of different modalities, for example Figure 4 As shown, the user turns the page of the webpage, and the new media content displayed during the page turning process can be checked for forgery. At least one modality of media content can appear on the same webpage, so the media content of any modality appearing on the same webpage can be checked for forgery at the same time or successively, and the corresponding areas can be marked respectively to prompt the first user that the media content in the corresponding area may be forged.

[0084] In some other implementations, the length of time that the first user stays on the real-time web page can also be detected. If the length of time is greater than a first preset length of time, at least one type of modal data in the real-time web page is detected for forgery, and a corresponding mark is set for the area that meets the preset forgery condition.

[0085] In some embodiments, the method further comprises:

[0086] Perform forgery detection on media content generated in at least two modes: a web page browsed in real time, an offline video played by the first terminal, an online video played in real time by the first terminal, a voice conversation conducted in real time between the first terminal and the second terminal, and a video conversation conducted in real time between the first terminal and the second terminal;

[0087] The detected forgery trace points are marked dynamically in real time.

[0088] It can be seen that by adopting the method provided by this embodiment, offline forgery detection and real-time online forgery detection of media content in diversified scenarios can be realized, and it can be dynamically changed based on different application scenarios generated by media content, which is suitable for multimodal media content detection. For users, both offline forgery detection and online real-time forgery detection can be realized. The combination of the two methods can provide detection information to users in a timely manner, remind users, and reduce users from being deceived or guided by forged generative artificial intelligence content. At the same time, it also meets the user's demand for forgery detection of the massive amount of media content received every day, and no longer needs to rely on third-party detection platforms or authoritative detection agencies, and the user experience is extremely high.

[0089] Figures 1 to 4 Any technical feature mentioned in any embodiment corresponding to any one of the items is also applicable to the embodiments of the present application. Figures 5 to 7 The corresponding embodiments will not be described in detail later.

[0090] The above describes a generative artificial intelligence content detection method in an embodiment of the present application. The following introduces a generative artificial intelligence content detection device that executes the above-mentioned generative artificial intelligence content detection method.

[0091] See also Figure 5 ,like Figure 5 The structure diagram of a generative artificial intelligence content detection device 40 shown in the figure can be applied to detect forgery of offline media content of any modality, online media content of any module, online voice conversation, online video conversation, etc., and prompt the user. The generative artificial intelligence content detection device 40 in the embodiment of the present application can realize the corresponding Figure 1-Figure 4 The steps in the generative artificial intelligence content detection method performed by the generative artificial intelligence content detection device 40 in any corresponding embodiment. The functions implemented by the generative artificial intelligence content detection device 40 can be implemented by hardware, or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions, and the modules can be software and / or hardware. The generative artificial intelligence content detection device 40 may include an input-output module 401 and a processing module 402. The functional implementation of the input-output module 401 and the processing module 402 can refer to Figure 1-Figure 4 The operations performed in any corresponding embodiment are not described in detail here.

[0092] In some implementations, the input-output module 401 may be used to obtain a first input from a first user on the detection interface, where the first input indicates content to be detected, and the content to be detected includes data in at least one state of text, image, video, or web page;

[0093] The processing module 402 is used to respond to the first input, determine a target detection mode according to the media type corresponding to the content to be detected, and perform forgery detection on the content to be detected in the target detection mode;

[0094] The processing module 402 is also used to generate an authentication result and output a first prompt information indicating the authentication result through the input-output module 401 if a target element that meets a preset forgery condition is detected in the content to be detected, wherein the authentication result indicates that at least one target element is forged content, and the preset forgery condition is data generated by at least one of generating an adversarial network, generating a diffusion model, or generating a large language model.

[0095] In some implementations, the first terminal runs an interactive application, the first user is in a real-time target conversation with a second user of at least one second terminal in a target mode, and the target mode includes a voice conversation or a video conversation; the processing module 402 is further configured to:

[0096] When it is detected that the first terminal starts the interactive application and the first user and the second user are in a real-time target session in a target mode;

[0097] Performing forgery detection on the real-time voice of at least one of the second users in the voice conversation, or performing forgery detection on the video character image of at least one of the second users in the video conversation;

[0098] If it is detected that at least one of the second users in the voice conversation meets the preset forgery condition, or if it is detected that at least one of the second users in the video conversation meets the preset forgery condition, a first mark is set in the voice conversation interface or the video conversation interface, and the first mark is used to indicate the confidence level that the second user is not the real person.

[0099] In some implementations, the processing module 402 is further configured to:

[0100] If in the real-time target session, there are a second user and a non-second user and both generate corresponding media content in the real-time target session, then after detecting the presence of voice that does not belong to the real second user in the real-time target session, the duration and appearance time of the media content corresponding to the second user and the non-second user are analyzed and marked, and the second prompt information is output through the input-output module 401.

[0101] In some implementations, the processing module 402 is specifically configured to:

[0102] Acquire first historical data of the second user, where the first historical data is obtained based on voice data of the second user in a voice conversation between the first user and the second user within a historical period;

[0103] According to the first historical data and the preset voice detection mode, forgery detection is performed on the real-time voice of at least one of the second users in the voice conversation.

[0104] In some implementations, the processing module 402 is specifically configured to:

[0105] Acquire second historical data of the second user, where the second historical data is obtained based on video data of the second user in a video conversation between the first user and the second user within a historical period;

[0106] According to the first historical data and the preset video detection mode, forgery detection is performed on the real-time video picture of at least one of the second users in the video conversation.

[0107] In some implementation modes, the first terminal displays the real-time web page browsed by the first user, and the processing module 402 is further configured to:

[0108] Following the changes in the real-time webpage content displayed when the first user browses the real-time webpage, detecting in real time a target area in the real-time webpage content that meets the preset forgery condition;

[0109] Dynamically setting a corresponding second mark for at least one target area detected in real time;

[0110] The second mark includes at least one of the following items:

[0111] A first confidence level indicating that the text in the target area meets a preset forgery condition;

[0112] A second confidence level indicating that the image in the target area meets a preset forgery condition;

[0113] Alternatively, a third confidence level indicating that the video played in the target area meets a preset forgery condition.

[0114] In some implementations, the processing module 402 is further configured to:

[0115] Perform forgery detection on media content generated in at least two modes: a web page browsed in real time, an offline video played by the first terminal, an online video played in real time by the first terminal, a voice conversation conducted in real time between the first terminal and the second terminal, and a video conversation conducted in real time between the first terminal and the second terminal;

[0116] The detected forgery trace points are marked dynamically in real time.

[0117] In the embodiment of the present application, by adopting the method provided in this embodiment, it is possible to realize offline forgery detection and real-time online forgery detection of media content in diversified scenarios, which can be dynamically changed based on different application scenarios generated by media content, and is suitable for multimodal media content detection. For users, both offline forgery detection and online real-time forgery detection can be realized. The combination of the two methods can provide detection information to users in a timely manner, remind users, and reduce users from being deceived or guided by forged generative artificial intelligence content. At the same time, it also meets the user's demand for forgery detection of the massive amount of media content received every day, and no longer needs to rely on third-party detection platforms or authoritative detection agencies, and the user experience is extremely high.

[0118] The above describes the generative artificial intelligence content detection device 40 for executing the generative artificial intelligence content detection method in the embodiment of the present application from the perspective of modular functional entities. The following describes the generative artificial intelligence content detection device 40 for executing the generative artificial intelligence content detection method in the embodiment of the present application from the perspective of hardware processing. It should be noted that in the embodiment of the present application Figure 7 The entity device corresponding to the input / output module 401 in the illustrated embodiment may be an input / output unit, a transceiver, a radio frequency circuit, a communication module, an output interface, etc., and the entity device corresponding to the processing module 402 may be a processor. Figure 7 The generative artificial intelligence content detection device 40 shown may have the following Figure 6 The structure shown, when Figure 5 The generative artificial intelligence content detection device 40 shown has the following features: Figure 6 When the structure shown is Figure 6 The processor and transceiver in the device can realize the same or similar functions as the input-output module 401 and the processing module 402 provided in the device embodiment corresponding to the generative artificial intelligence content detection device 40. Figure 6 The memory in the storage processor is a computer program that needs to be called when executing the above-mentioned generative artificial intelligence content detection method.

[0119] The present application also provides another generative artificial intelligence content detection device, such as Figure 7As shown, for the sake of convenience, only the parts related to the embodiments of the present application are shown. For specific technical details not disclosed, please refer to the method part of the embodiments of the present application. The generative artificial intelligence content detection device can be any generative artificial intelligence content detection device including a mobile phone, a tablet computer, a personal digital assistant (English full name: Personal Digital Assistant, English abbreviation: PDA), a sales generative artificial intelligence content detection device (English full name: Point of Sales, English abbreviation: POS), a car computer, etc., taking the generative artificial intelligence content detection device as a mobile phone as an example:

[0120] Figure 7 The block diagram shows a partial structure of a mobile phone related to the generative artificial intelligence content detection device provided in the embodiment of the present application. Figure 7 The mobile phone includes: a radio frequency (RF) circuit 710, a memory 720, an input unit 730, a display unit 740, a sensor 780, an audio circuit 760, a wireless fidelity (Wi-Fi) module 7100, a processor 780, and a power supply 790. Those skilled in the art will appreciate that Figure 7 The mobile phone structure shown in the figure does not constitute a limitation on the mobile phone, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0121] Combine the following Figure 7 A detailed introduction to the various components of the mobile phone:

[0122] The RF circuit 710 can be used for receiving and sending signals during information transmission or calls. In particular, after receiving the downlink information of the base station, it is sent to the processor 780 for processing; in addition, the designed uplink data is sent to the base station. Generally, the RF circuit 710 includes but is not limited to an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier (full name: LowNoiseAmplifier, English abbreviation: LNA), a duplexer, etc. In addition, the RF circuit 710 can also communicate with the network and other devices through wireless communication. The above-mentioned wireless communications may use any communication standard or protocol, including but not limited to Global System of Mobile communications (Global System of Mobile communication, English abbreviation: GSM), General Packet Radio Service (General Packet Radio Service, English abbreviation: GPRS), Code Division Multiple Access (Code Division Multiple Access, English abbreviation: CDMA), Wideband Code Division Multiple Access (WCDMA), Long Term Evolution (Long Term Evolution, English abbreviation: LTE), e-mail, Short Messaging Service (SMS), etc.

[0123] The memory 720 can be used to store software programs and modules. The processor 780 executes various functional applications and data processing of the mobile phone by running the software programs and modules stored in the memory 720. The memory 720 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory 720 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0124] The input unit 730 can be used to receive input digital or character information, and to generate key signal input related to the user settings and function control of the mobile phone. Specifically, the input unit 730 may include a touch panel 731 and other input devices 732. The touch panel 731, also known as a touch screen, can collect the user's touch operation on or near it (such as the user's operation on the touch panel 731 or near the touch panel 731 using any suitable object or accessory such as a finger, stylus, etc.), and drive the corresponding connection device according to a pre-set program. Optionally, the touch panel 731 may include two parts: a touch detection device and a touch controller. Among them, the touch detection device detects the user's touch orientation, detects the signal brought by the touch operation, and transmits the signal to the touch controller; the touch controller receives the touch information from the touch detection device, converts it into contact coordinates, and then sends it to the processor 780, and can receive and execute commands sent by the processor 780. In addition, the touch panel 731 can be implemented in various types such as resistive, capacitive, infrared, and surface acoustic waves. In addition to the touch panel 731, the input unit 730 may further include other input devices 732. Specifically, the other input devices 732 may include but are not limited to one or more of a physical keyboard, function keys (such as volume control keys, switch keys, etc.), a trackball, a mouse, a joystick, and the like.

[0125] The display unit 740 can be used to display information input by the user or information provided to the user and various menus of the mobile phone. The display unit 740 may include a display panel 741. Optionally, the display panel 741 may be configured in the form of a liquid crystal display (full name in English: Liquid Crystal Display, English abbreviation: LCD), an organic light-emitting diode (full name in English: Organic Light-Emitting Diode, English abbreviation: OLED), etc. Further, the touch panel 731 may cover the display panel 741. When the touch panel 731 detects a touch operation on or near it, it is transmitted to the processor 780 to determine the type of touch event. Subsequently, the processor 780 provides a corresponding visual output on the display panel 741 according to the type of touch event. Although in Figure 7 In the embodiment, the touch panel 731 and the display panel 741 are used as two independent components to realize the input and output functions of the mobile phone, but in some embodiments, the touch panel 731 and the display panel 741 can be integrated to realize the input and output functions of the mobile phone.

[0126] The mobile phone may also include at least one sensor 780, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor, wherein the ambient light sensor may adjust the brightness of the display panel 741 according to the brightness of the ambient light, and the proximity sensor may turn off the display panel 741 and / or the backlight when the mobile phone is moved to the ear. As a type of motion sensor, the accelerometer sensor can detect the magnitude of acceleration in all directions (generally three axes), and can detect the magnitude and direction of gravity when stationary. It can be used for applications that identify the posture of the mobile phone (such as horizontal and vertical screen switching, related games, magnetometer posture calibration), vibration recognition related functions (such as pedometer, tapping), etc.; as for other sensors that can be configured in the mobile phone, such as gyroscopes, barometers, hygrometers, thermometers, infrared sensors, etc., they will not be repeated here.

[0127] The audio circuit 760, the speaker 761, and the microphone 762 can provide an audio interface between the user and the mobile phone. The audio circuit 760 can transmit the received audio data to the speaker 761 after converting the received audio data into an electrical signal, which is converted into a sound signal for output; on the other hand, the microphone 762 converts the collected sound signal into an electrical signal, which is received by the audio circuit 760 and converted into audio data, and then the audio data is output to the processor 780 for processing, and then sent to another mobile phone through the RF circuit 710, or the audio data is output to the memory 720 for further processing.

[0128] Wi-Fi is a short-range wireless transmission technology. Mobile phones can help users send and receive emails, browse web pages, and access streaming media through the Wi-Fi module 7100. It provides users with wireless broadband Internet access. Figure 6 A Wi-Fi module 7100 is shown, but it is understandable that it is not an essential component of the mobile phone and can be omitted as needed without changing the essence of the application.

[0129] The processor 780 is the control center of the mobile phone. It uses various interfaces and lines to connect various parts of the entire mobile phone. By running or executing software programs and / or modules stored in the memory 720, and calling data stored in the memory 720, it executes various functions of the mobile phone and processes data, thereby monitoring the mobile phone as a whole. Optionally, the processor 780 may include one or more processing units; preferably, the processor 780 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 780.

[0130] The mobile phone also includes a power supply 790 (such as a battery) for supplying power to various components. The power supply can be logically connected to the processor 780 through a power management system, so that the power management system can manage charging, discharging, power consumption and other functions.

[0131] Although not shown, the mobile phone may also include a camera, a Bluetooth module, etc., which will not be described in detail here.

[0132] In the embodiment of the present application, the processor 780 included in the mobile phone also has the function of controlling and executing the above Figure 7 The steps performed by the generative artificial intelligence content detection device 40 in the above embodiment can be based on the Figure 7 For example, the processor 722 performs the following operations by calling the instructions in the memory 732:

[0133] Acquiring, through the input unit 730, a first input of a first user on the detection interface, where the first input indicates content to be detected, and the content to be detected includes data in at least one state of text, image, video, or web page;

[0134] In response to the first input, determining a target detection mode according to a media type corresponding to the content to be detected, and performing forgery detection on the content to be detected under the target detection mode;

[0135] If it is detected that there is a target element that meets the preset forgery condition in the content to be detected, a forgery detection result is generated and a first prompt information indicating the forgery detection result is output through the input unit 730, and the forgery detection result indicates that at least one target element is forged content, and the preset forgery condition is data generated by at least one of generation based on a generative adversarial network, generation based on a diffusion model, or generation based on a large language model.

[0136] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0137] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0138] In the several embodiments provided in the embodiments of the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules, which can be electrical, mechanical or other forms.

[0139] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0140] In addition, each functional module in each embodiment of the present application can be integrated into a processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The above-mentioned integrated module can be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0141] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0142] The computer program product includes one or more computer instructions. When the computer program is loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium, or transmitted from a computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website site, a computer, a server, or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server, or data center. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a server or a data center that includes one or more available media integrations. The available medium may be a magnetic medium, (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk Solid State Disk (SSD)), etc.

[0143] The technical solutions provided in the embodiments of the present application are introduced in detail above. The principles and implementation methods of the embodiments of the present application are explained by using specific examples in the embodiments of the present application. The description of the above embodiments is only used to help understand the methods and core ideas of the embodiments of the present application. At the same time, for those skilled in the art, according to the ideas of the embodiments of the present application, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on the embodiments of the present application.

Claims

1. A generative artificial intelligence content detection method, characterized in that: The method is applied to a first terminal, the first terminal displays a detection interface, and the method includes: Acquire a first input of a first user in the detection interface, where the first input indicates content to be detected, and the content to be detected includes data in at least one state of text, image, video, or web page; In response to the first input, determining a target detection mode according to a media type corresponding to the content to be detected, and performing forgery detection on the content to be detected under the target detection mode; If it is detected that there is a target element that meets the preset forgery condition in the content to be detected, an authentication result is generated and a first prompt information indicating the authentication result is output, and the authentication result indicates that at least one target element is forged content, and the preset forgery condition is data generated based on at least one of a generative adversarial network, a diffusion model, or a large language model.

2. The method according to claim 1, characterized in that The first user and the second user of at least one second terminal are in a real-time target conversation in a target mode, wherein the target mode includes a voice conversation or a video conversation; and the method further includes: When it is detected that the first terminal starts the interactive application and the first user and the second user are in a real-time target session in a target mode; Performing forgery detection on the real-time voice of at least one of the second users in the voice conversation, or performing forgery detection on the video character image of at least one of the second users in the video conversation; If it is detected that at least one of the second users in the voice conversation meets the preset forgery condition, or if it is detected that at least one of the second users in the video conversation meets the preset forgery condition, a first mark is set in the voice conversation interface or the video conversation interface, and the first mark is used to indicate the confidence level that the second user is not the real person.

3. The method according to claim 2, characterized in that The method further comprises: If in the real-time target session, there are a second user and a non-second user and both generate corresponding media content in the real-time target session, then after detecting the presence of voice that does not belong to the real second user in the real-time target session, the duration and appearance time of the media content corresponding to the second user and the non-second user are analyzed and marked, and a second prompt information is output.

4. The method according to claim 2 or 3, characterized in that: The performing forgery detection on the real-time voice of at least one of the second users in the voice conversation includes: Acquire first historical data of the second user, where the first historical data is obtained based on voice data of the second user in a voice conversation between the first user and the second user within a historical period; According to the first historical data and the preset voice detection mode, forgery detection is performed on the real-time voice of at least one of the second users in the voice conversation.

5. The method according to claim 2 or 3, characterized in that: The performing forgery detection on the video character image of at least one of the second users in the video session includes: Acquire second historical data of the second user, where the second historical data is obtained based on video data of the second user in a video conversation between the first user and the second user within a historical period; According to the first historical data and the preset video detection mode, forgery detection is performed on the real-time video picture of at least one of the second users in the video conversation.

6. The method according to claim 1, characterized in that The first terminal displays the real-time web page browsed by the first user, and the method further includes: Following the changes in the real-time webpage content displayed when the first user browses the real-time webpage, detecting in real time a target area in the real-time webpage content that meets the preset forgery condition; Dynamically setting a corresponding second mark for at least one target area detected in real time; The second mark includes at least one of the following items: A first confidence level indicating that the text in the target area meets a preset forgery condition; A second confidence level indicating that the image in the target area meets a preset forgery condition; Alternatively, a third confidence level indicating that the video played in the target area meets a preset forgery condition.

7. The method according to any one of claims 1 to 4 and 6, characterized in that: The method further comprises: Perform forgery detection on media content generated in at least two modes: a web page browsed in real time, an offline video played by the first terminal, an online video played in real time by the first terminal, a voice conversation conducted in real time between the first terminal and the second terminal, and a video conversation conducted in real time between the first terminal and the second terminal; The detected forgery trace points are marked dynamically in real time.

8. A computer device, characterized in that: The computer device comprises: at least one processor and memory; The memory is used to store a computer program, and the processor is used to call the computer program stored in the memory to execute the method according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that: The method comprises instructions which, when executed on a computer, cause the computer to execute the method according to any one of claims 1 to 7.

10. A computer program product comprising instructions, characterized in that When the method is executed on a computer, the computer is enabled to execute the method according to any one of claims 1 to 7.

Images