Generative artificial intelligence content processing method and device in interaction and storage medium

Abstract

The invention relates to the field of AIGC, and provides a generative artificial intelligence content processing method in interaction, a related device and a storage medium, and the method comprises the steps: a content detection tool operates in a plug-in mode, and detects a first touch input of a user for a first region of a first display interface, the first area comprises at least one candidate element of at least one mode to be detected by a user; in response to the first touch input, scanning the first area to obtain candidate elements and displaying the candidate elements in the first window; in response to a second touch input for the candidate elements, skipping to a second display interface, and performing content detection on the candidate elements by using a target detection mode matched with the modality of each candidate element; and setting an annotation for the detected first target element of the AIGC content, wherein the annotation is used for displaying the confidence coefficient, the generation mode and the data source that the first target element is the AIGC content. According to the scheme, a terminal user can conveniently and quickly know the generation modes of the interactive messages of different modes, coherent tracing is facilitated, and the terminal user is prevented from being cheated.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of generative artificial intelligence technology, and in particular to a method, device and storage medium for processing generative artificial intelligence content in interaction. 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 produced by generative artificial intelligence tools is difficult to distinguish between true and false, spreads rapidly, and is seriously misleading to 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 generative artificial intelligence (AIGC, Artificial Intelligence Generated Content), especially the forgery of text and audio is increasing, and 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 they cannot be distinguished manually, which brings great obstacles to the identification of judicial evidence. In addition, the current products can only detect forgery of offline media content, and are also limited to the server side. Now users are more inclined to read various media content on the terminal, and conduct voice or video conversations, but can only analyze forgery behavior afterwards. Usually, various interactive messages based on social software such as interactive applications cannot be directly AIGC detected afterwards and the detection results cannot be displayed on the terminal side for users to view at any time, nor can AIGC detection be performed in real time during interaction and the detection results can be displayed in real time on the terminal side to remind users participating in the interaction. Therefore, it is not possible to directly perform intuitive AIGC detection on the terminal side for the session generating interactive messages and display the detection results in real time, resulting in users not being able to distinguish the forgery in the interactive messages in the first time. Summary of the invention

[0004] The embodiments of the present application provide a generative artificial intelligence content processing method, related devices and storage media, which can improve the efficiency of the generative artificial intelligence content detection method, and can perform multi-dimensional and real-time authentication of the content to be detected (such as interactive messages generated in the interaction) in a variety of mainstream application scenarios for terminal users, so that the authentication results are more real-time, 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, so that terminal users can 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 the first aspect, the embodiment of the present application introduces a generative artificial intelligence content processing method from the terminal side, the method is executed by a content detection tool, and the content detection tool runs in at least one of the following ways: in the terminal background, inserted into a web page, or embedded in an input method in a plug-in manner; the method includes:

[0006] Acquire a first touch input of a user on a first area of ​​a first display interface, where the first area includes at least one candidate element of at least one modality to be detected by the user;

[0007] In response to the first touch input, scanning the first area based on machine vision technology to obtain at least one candidate element of at least one modality and display it in a pop-up first window;

[0008] Acquire a second touch input of a user for any candidate element in the first window;

[0009] In response to the second touch input, jump to the second display interface of the content detection tool, and select a target detection mode adapted to the modality to perform content detection on at least one of the candidate elements according to the modality of each candidate element in the first window;

[0010] If it is detected that the first area includes a first target element that conforms to at least one preset generative artificial intelligence content generation method, a first annotation is set and displayed for the first target element, and the first annotation is used to display the confidence, generation method and data source that the first target element is generative artificial intelligence content.

[0011] In a second aspect, an embodiment of the present application provides a content detection device having the function of implementing the interactive generative artificial intelligence content processing 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.

[0012] In some implementations, the content detection tool runs in at least one of a plug-in in the terminal background, inserted into a web page, or embedded in an input method; the content detection device includes an input and output module, a processing module, and a display module.

[0013] The input-output module is configured to obtain a first touch input from a user to a first area of ​​a first display interface, where the first area includes at least one candidate element of at least one modality to be detected by the user;

[0014] The processing module is configured to scan the first area based on machine vision technology in response to the first touch input acquired by the input-output module, obtain at least one candidate element of at least one modality, and display the candidate element in a pop-up first window through the display module;

[0015] The input-output module is further used to obtain a second touch input from a user for any candidate element in the first window;

[0016] The processing module is further configured to, in response to the second touch input acquired by the input / output module, control the display module to jump to the second display interface of the content detection tool, and select a target detection method adapted to the modality to perform content detection on at least one of the candidate elements according to the modality of each candidate element in the first window;

[0017] If it is detected that the first area includes a first target element that conforms to at least one preset generative artificial intelligence content generation method, a first annotation is set for the first target element and displayed through the display module, and the first annotation is used to display the confidence, generation method and data source of the first target element as generative artificial intelligence content.

[0018] In some implementations, the processing module is specifically used to

[0019] In response to the second touch input, popping up a new window for pattern recognition through the display module;

[0020] In response to a user clicking an input on a generative artificial intelligence AIGC content detection function icon in the new window, performing content detection on the selected candidate element;

[0021] Displaying the obtained detection result in the new window through the display module;

[0022] The detection result is associated with a first target element that conforms to at least one preset generative artificial intelligence content generation method in a first annotation manner; or after hiding the new window, the detection result is associated and annotated with the first annotation in the area corresponding to the first target element in the original text to which the first target element belongs.

[0023] In some implementations, the method is applied to detect data of any mode appearing in any manner in an instant messaging application; the first display interface is a session interface in which the user is currently participating, and the session interface is a single session interface or a group session interface; the processing module is further used to:

[0024] Acquire historical conversation elements of any mode generated in the first display interface within a target time period, wherein the conversation elements include at least one of text, document, picture, video and voice;

[0025] According to the modality of the historical conversation elements, selecting a target detection method adapted to the modality to perform content detection on at least one of the historical conversation elements of the at least one modality;

[0026] If it is detected that there is a second target element in the historical conversation element that conforms to at least one preset generative artificial intelligence content generation method, a second annotation is set for the second target element and displayed on the first display interface through the display module, and the second annotation is used to display the confidence, generation method and data source of the second target element as generative artificial intelligence content.

[0027] In some implementations, the processing module is specifically used to:

[0028] After determining that there is an interactive message matching a preset keyword in the conversation interface within the target time period, obtaining a candidate element of any modality matching the preset keyword in the interactive message, where the interactive message is a document, a picture, or a video;

[0029] According to the mode of the interactive message, the content of each candidate element is detected;

[0030] The detection results are annotated and displayed on the corresponding interactive message.

[0031] In some implementations, the method is applied to a news and document retrieval system displayed in real time on a terminal, and the processing module is further used to:

[0032] Detecting, based on the input-output module, a third touch input of the user on the conversation interface to the target interactive message;

[0033] In response to the third touch input, displaying a content detection window of the target interactive message;

[0034] In response to the user selecting the target content in the content detection window, the display module jumps to the detection page of the content detection tool, displays the corresponding detection analysis report, and annotates the summary of the detection analysis report and displays it on the corresponding interactive message.

[0035] In some implementations, the first target element is a picture that meets a preset business requirement, and the processing module is further configured to:

[0036] Obtaining content description information of the first target element;

[0037] According to the content description information of the preset business requirement, obtaining a preset confrontation pattern matching the content description information from a preset confrontation pattern library;

[0038] The first target element and the preset adversarial pattern are synthesized, and at least one adversarial sample obtained by the synthesis is updated into an adversarial sample library, where the adversarial sample is used for adversarial training.

[0039] In some embodiments, the method further comprises:

[0040] Starting a camera in communication with the terminal to detect a first gaze area of ​​a user facing the first display interface;

[0041] If the first gaze area includes candidate elements that conform to a preset modality, selecting a target detection method adapted to the modality to perform content detection on at least one candidate element in the first gaze area according to the modality of each candidate element in the first gaze area;

[0042] If it is detected that the first gaze area includes a third target element that conforms to at least one preset generative artificial intelligence content generation method, a third annotation is set and displayed for the third target element, and the third annotation is used to display the confidence, generation method and data source of the third target element as generative artificial intelligence content.

[0043] In some implementations, the method is applied to a virtual reality scenario, the user wears a virtual reality helmet, and the processing module is further used to:

[0044] Track and detect the user's gaze in real time, and dynamically lock the second gaze area corresponding to the user's gaze;

[0045] If it is detected that the second gaze area includes a fourth target element that conforms to at least one preset generative artificial intelligence content generation method, a fourth annotation is set and displayed for the third target element, and the fourth annotation is used to display the confidence, generation method and data source of the fourth target element as generative artificial intelligence content.

[0046] In a fifth aspect, an embodiment of the present application provides a content detection device, comprising: 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 processing methods provided in the second aspect.

[0047] In a sixth aspect, an embodiment of the present application provides a computer-readable storage medium having a function of implementing the generative artificial intelligence content processing 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 the processor to load to execute the steps in the first aspect, any one of the implementation methods of the first aspect, or any one of the generative artificial intelligence content processing methods provided in the second aspect in the embodiment of the present application.

[0048] Compared with the prior art, in the solution provided by the embodiment of the present application, since the content detection tool runs in the terminal background, inserted into a web page, or implanted in at least one way in the input method in a plug-in manner, when the user uses the built-in interactive application of the terminal, the interactive message in the interactive application can be automatically detected, and the first target element detected that conforms to at least one preset generative artificial intelligence content generation method is marked and displayed in time. Alternatively, after receiving the first touch input initiated by the user to the first display interface, the candidate elements of various modes in the first area are scanned and identified and displayed in the first pop-up window, so that the user can select any candidate element that he wants to detect, and select the target detection method adapted to its modality for the candidate element selected by the user for content detection, and timely set the first mark and display the first target element detected that conforms to at least one preset generative artificial intelligence content generation method. It can be seen that on the one hand, it allows users to intuitively see which interactive messages sent by other users participating in the interaction are generated based on AIGC in real-time interaction, which is convenient for users to quickly understand, thereby reducing the risk of misidentification and having a certain understanding of the credibility of interactive messages. On the other hand, it allows users to quickly understand the authenticity of interactive messages intuitively without having to look them up on their own or having to go through relevant knowledge thresholds. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A schematic diagram of an application environment provided in an embodiment of the present application;

[0050] Figure 2 A schematic diagram of a flow chart of a generative artificial intelligence content processing method in an embodiment of the present application;

[0051] Figure 3 A schematic diagram of a flow chart of a generative artificial intelligence content processing method in an embodiment of the present application;

[0052] Figure 4a A schematic diagram of an interface change for detecting whether a document in a group message is AIGC content in an embodiment of the present application;

[0053] Figure 4b A schematic diagram of an interface change for detecting whether a picture in a group message is AIGC content in an embodiment of the present application;

[0054] Figure 4c This is a schematic diagram of an interface for marking AIGC content in a group message in an embodiment of the present application;

[0055] Figure 4d This is a schematic diagram of an interface for marking voice and video generated based on AIGC in a group message in an embodiment of the present application;

[0056] Figure 5 A schematic diagram of an interface change for detecting whether a document in a group message is AIGC content in an embodiment of the present application;

[0057] Figure 6 A schematic diagram of detecting AIGC content in a group message based on intelligent eye tracking in an embodiment of the present application;

[0058] Figure 7 is a structural diagram of a content detection device in an embodiment of the present application;

[0059] Figure 8 It is a structural schematic diagram of a physical device for implementing the interactive generative artificial intelligence content processing method in the embodiment of the present application;

[0060] Fig. 9 It is a structural diagram of a mobile phone that implements the interactive generative artificial intelligence content processing method in the embodiment of the present application. DETAILED DESCRIPTION

[0061] The terms "first", "second", etc. in the description and claims of the embodiments of the present application and the above-mentioned drawings are used to distinguish similar objects (for example, the first target element and the second target element in the embodiments of the present application respectively represent different elements with the same attributes), and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. 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 that are 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. In addition, the modules or submodules 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 can be selected according to actual needs to achieve the purpose of the embodiment of the present application.

[0062] The embodiments of the present application provide a method for processing generative artificial intelligence content in interaction, related devices and storage media, which can be used in a server or terminal device. Specifically, it can be used to authenticate whether interactive messages appearing in mainstream interactive applications are generative artificial intelligence content, so as to distinguish them from manually written texts, real images and videos taken by a shooting device, and real sounds spoken by real humans. It can also authenticate the media content (such as sound and facial images) generated in real-time voice conversations and video conversations.

[0063] In some implementations, the present invention is applied to Figure 1 The communication system shown in FIG. 1 may include at least one service server, at least two terminals ( Figure 1Only 2 are marked). Among them, the business server and each terminal are connected by communication, and the at least two terminals may be at least one terminal bound to a user account with an associated relationship with each other. The association relationship may include a friend relationship between two user accounts, or the two user accounts may belong to the same interactive group (also called a session group, without limiting the name). The interactive group may be an interactive application of the same type (such as WeChat APP), or it may be an interactive application of different types (such as WeChat APP, enterprise WeChat APP) that has established a cross-domain communication connection. This application does not limit the type of interaction and the association between the user accounts participating in the interaction, as long as they can interact with each other in a one-to-one or one-to-many interaction scenario.

[0064] A content detection tool may be installed in the terminal, and the content detection tool may be run in at least one of the following ways: in the terminal background in the form of a plug-in, inserted into a web page, or embedded in an input method.

[0065] In some implementations, the content detection tool may include the following functional modules:

[0066] 1) Detection content input module: This module supports the input of video, image, audio and text file types. The user can touch any interactive message that appears in the interactive interface so that the interactive message is scanned to obtain at least one candidate element of at least one modality. The user can then touch any candidate element to select the candidate element that he wants to authenticate.

[0067] 2) Algorithm detection module: The algorithm detection link is to select the target detection method adapted to the modality of the candidate elements selected by the user to perform content detection on at least one of the candidate elements. After the algorithm reasoning operation is completed, the detection results will be output for the user to view.

[0068] 3) Result display module: After the detection is completed, the detection results will be output. 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 can view the detection report offline or online at any time.

[0069] The generative AI content detection device can detect forgery of interactive messages (such as text, video, image or audio) that are circulated or appear in interactive applications used by users, and can also detect forgery of media content generated in communication sessions (such as 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 personal reputation infringement, combating network fraud, and improving the security of biometric systems.

[0070] 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:

[0071] 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.

[0072] 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.

[0073] 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.

[0074] Aiming at the face recognition direction in the field of artificial intelligence, the present application can use a face image set to generate a universal perturbation with high attack robustness, and by adding the universal perturbation to the face images in the face image set, the image classification model cannot accurately recognize the face images with the added perturbation, thereby realizing the attack on the image classification model.

[0075] It should be noted that the terminals and content detection tools involved in the embodiments of the present application may be smart phones, tablet computers, laptop computers, desktop computers, smart speakers, smart watches, personal digital assistants, etc., but are not limited thereto. The terminals may be directly or indirectly connected via wired or wireless communication, which is not limited in the embodiments of the present application.

[0076] Regarding various AIGC abuse scenarios based on interactive applications, users now prefer to read various media content on the terminal, as well as conduct voice or video conversations. However, they can only analyze the forgery behavior afterwards. Usually, various interactive messages based on social software such as interactive applications cannot be directly AIGC-detected afterwards and the detection results cannot be displayed on the terminal side for users to view at any time. It is also impossible to perform real-time AIGC detection during the interaction and display the detection results on the terminal side in real time to remind users participating in the interaction. Therefore, it is not possible to directly perform intuitive AIGC detection on the terminal side for the conversation that generates the interactive message and display the detection results in real time, resulting in the user being unable to distinguish the forgery in the interactive message in the first time. In this regard, the embodiments of the present application are combined with Figure 2-Figure 9 The technical solution of the embodiment of the present application is exemplified.

[0077] like Figure 2 As shown, Figure 2 The present invention is a flow chart of a method for processing interactive generative artificial intelligence content, which is executed by a content detection tool, which is run in at least one of the following ways: in the terminal background, inserted into a web page, or embedded in an input method in the form of a plug-in. The method includes:

[0078] 101. Obtain a first touch input of a user on a first area of ​​a first display interface.

[0079] The first area includes at least one candidate element of at least one modality to be detected by the user.

[0080] In an embodiment of the present application, the first display interface may be any interface of an instant messaging application (such as a conversation interface, an article reading interface, a video playback interface, etc., without limitation to this), a streaming media playback application (such as a short video APP, a video playback platform, etc.). In an embodiment of the present application, when the first display interface is a conversation interface of an instant messaging application (taking a group conversation as an example, a single conversation may refer to an embodiment of a group conversation), it refers to content detection of text paragraphs, pictures, videos, audio or files in the chat window. The specific application scenario is not limited in this application, and you may refer to this scenario without further elaboration.

[0081] 102. In response to the first touch input, scan the first area based on machine vision technology to obtain at least one candidate element of at least one modality and display it in a pop-up first window.

[0082] For example, by scanning the first area based on a machine vision recognition plug-in built into the terminal, it can be determined whether the first area includes candidate elements of any modality, such as pictures, files, etc.

[0083] 103. Obtain a second touch input of the user for any candidate element in the first window.

[0084] 104. In response to the second touch input, jump to the second display interface of the content detection tool, and select a target detection method adapted to the modality to perform content detection on at least one of the candidate elements according to the modality of each candidate element in the first window.

[0085] Specifically, after detecting the user's second touch input for any target element in the first window, in response to the second touch input, jump to the second display interface of the content detection tool, and in response to the user's third touch input for the target function icon, according to the modality of each target element in the first window, select a target detection method adapted to the modality to perform content detection on at least one of the target elements. Among them, the target function icon is text detection, image detection, graphic and text detection, or video detection. For example, the detection function icon may pop up after the user selects the text that appears in the first window.

[0086] 105. If it is detected that the first area includes a first target element that conforms to at least one preset generative artificial intelligence content generation method, a first label is set for the first target element and displayed.

[0087] Among them, the first annotation is used to display the confidence, generation method and data source of the first target element as generative artificial intelligence content.

[0088] Specifically, in response to the second touch input, a new window for pattern recognition pops up; in response to the user clicking the AIGC content detection function icon in the new window, content detection is performed on the selected candidate element, and the obtained detection result is displayed in the new window;

[0089] The detection result is associated with a first target element that conforms to at least one preset generative artificial intelligence content generation method in a first annotation manner; or after hiding the new window, the detection result is associated and annotated with the first annotation in the area corresponding to the first target element in the original text to which the first target element belongs.

[0090] It is understandable that after the first target element is detected, it can be output in at least one of the following ways:

[0091] Display in the form of a prompt window;

[0092] Displayed in a marked box;

[0093] Display with preset annotation shapes;

[0094] Display as background message;

[0095] Or output in voice broadcast mode.

[0096] The embodiment itself does not limit the display method, and the display method can be customized by the user or preset by the content detection tool. For group conversations or single conversation scenarios, the interactive message corresponding to the first target element can be highlighted, and the highlighting method is not limited.

[0097] 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.

[0098] In the solution provided by the embodiment of the present application, since the content detection tool runs in the terminal background, inserted into a web page, or implanted in at least one way in the form of a plug-in, when the user uses the built-in interactive application of the terminal, the interactive message in the interactive application can be automatically detected, and the first target element detected that conforms to at least one preset generative artificial intelligence content generation method is marked and displayed in time. Or, after receiving the first touch input initiated by the user on the first display interface, the candidate elements of various modes in the first area are scanned and identified and displayed in the first pop-up window, so that the user can select any candidate element that he wants to detect, and the target detection method adapted to the modality of the candidate element selected by the user is selected for content detection, and the first target element detected that conforms to at least one preset generative artificial intelligence content generation method is set with a first mark and displayed in time. It can be seen that on the one hand, the user can intuitively see which interactive messages sent by other users participating in the interaction are generated based on AIGC in real-time interaction, which is convenient for the user to quickly understand, thereby reducing the risk of misidentification and having a certain understanding of the credibility of the interactive message. On the other hand, the user can intuitively understand the authenticity of the interactive message quickly without self-inquiry or relevant knowledge threshold.

[0099] Optionally, in some embodiments of the present application, the method can be applied to detect data of any modality appearing in any manner in an instant messaging application; accordingly, the first display interface is the session interface in which the user is currently participating, and the session interface is a single session interface or a group session interface. In order to facilitate users to trace back or review data of any modality appearing in historical sessions at any time in the era of information explosion, the present embodiment can also implement content detection of any historical session elements in the single session or group session in which the user is located, and partially or completely display the content detection results of the corresponding interactive messages. The content detection in the present application can be automatic (that is, as long as the terminal currently displays a session interface, the content detection results are automatically marked in the session interface), or it can be actively triggered by the user, mainly taking the user's active trigger as an example. Specifically, if Figure 3-4d As shown, the method also includes:

[0100] 301. Obtain historical conversation elements of any mode generated in the first display interface within a target time period.

[0101] The conversation element includes at least one of text, document, picture, video and voice.

[0102] 302. According to the modality of the historical conversation element, select a target detection method adapted to the modality to perform content detection on at least one historical conversation element of the at least one modality.

[0103] 303. If it is detected that a second target element that conforms to at least one preset generative artificial intelligence content generation method exists in the historical conversation element, a second annotation is set for the second target element.

[0104] Among them, the second annotation is used to display the confidence, generation method and data source of the second target element as generative artificial intelligence content.

[0105] In some implementations, the following steps may be used to perform content detection on the above historical conversation elements:

[0106] First, after determining that there is an interactive message matching a preset keyword in the conversation interface within the target time period, a candidate element of any modality matching the preset keyword in the interactive message is obtained, where the interactive message is a document, a picture, or a video;

[0107] Then, according to the mode of the interactive message, the content of each candidate element is detected;

[0108] The detection results are then associated and annotated and displayed on the corresponding interactive message.

[0109] For example, if Figure 4a As shown, in the conversation of group 1 (399 users) where user a is located, after user b sends a document named "XX video processing method" in group 1, the content detection result of the document can be displayed on the conversation interface of group 1 of user a, and the AIGC paragraphs detected in the document can be annotated (the detailed detection results corresponding to the annotation can be folded, and the user can expand them by clicking).

[0110] like Figure 4b As shown, after user b sends picture a in group 1, the content detection result of picture a can be displayed on the conversation interface of user a's group 1, for example, the local or global area in picture a that meets AIGC is marked.

[0111] like Figure 4c As shown, after user b sends a document named "XX video processing method" and picture a in group 1, the content detection results of the document and picture a (i.e., the marks indicated by the solid circles) are displayed in sequence on the conversation interface of user a's group 1.

[0112] like Figure 4d As shown, after user b sends a voice and a video in group 1, the content detection results of the voice and the video (ie, the marks indicated by the solid circles) are displayed in sequence on the conversation interface of user a's group 1.

[0113] It is understandable that user a refers to any user in group 1 who receives an interactive message sent by user b in group 1, and user b refers to any user who sends an interactive message in group 1. The terminal where user a is located may be installed with a related content detection tool that can implement the embodiment of the present application, or a certain version of an interactive application that can implement the embodiment of the present application. As for which method user a chooses to help himself realize intelligent content detection, this application does not limit it.

[0114] In other implementations, the marks may be divided into several levels according to the confidence level in the detection results, and different colors of marks may be used for different levels of detection results. This is not limited to this and is only used for distinction to remind users.

[0115] It can be seen that by conducting targeted content detection on interactive messages that match preset keywords, on the one hand, it is convenient for users to check the authenticity of interactive messages that they are more concerned about at any time; on the other hand, it is not necessary to conduct content detection on all interactive messages, thereby reducing the impact of the full screen display of content detection results on the user's browsing experience of the session. On the other hand, it allows users to quickly understand the authenticity of interactive messages intuitively without having to check them by themselves or having relevant knowledge thresholds, which greatly improves the user experience, especially in the era of information explosion.

[0116] Optionally, in some embodiments of the present application, the method may be applied to a news or document retrieval system displayed in real time on a terminal, and may annotate AIGC paragraphs in articles and local regions (AIGC) in images. Specifically, the method further includes:

[0117] Detecting a third touch input of the user on the conversation interface to the target interactive message;

[0118] In response to the third touch input, displaying a content detection window of the target interactive message;

[0119] In response to the user selecting the target content in the content detection window, jumping to the detection page of the content detection tool and displaying the corresponding detection analysis report;

[0120] The summary of the detection and analysis report is associated and marked and displayed on the corresponding interactive message, so that the user can check the authenticity of the corresponding interactive message at any time.

[0121] For example Figure 5As shown, user a clicks on the conversation interface of group 2, and user b sends a document b in group 2, and sends the corresponding text "Hello everyone, this is a relatively professional article about XX that I saw recently, sharing it with everyone." User a touches and stays on the document b (regardless of whether the document b is open) for a certain period of time, and jumps or directly pops up the content detection window for the document b, displaying the field content, content summary or paragraph number (i.e. Figure 5 Fig2 in Fig. 2 has a shaded rectangular area), user a selects the key field or paragraph number to be detected (such as Figure 5 Fig3 in the figure can detect the content and obtain the detection analysis report or report link corresponding to the key field or paragraph number ( Figure 5 Finally, the summary of the detection and analysis report generates a link and associates it with the document b displayed in the conversation interface of user a (such as Figure 5 Fig4 in the figure enables user a (or any user in group 2) to directly read the detection analysis report of document b by clicking the summary link of the associated annotation. It is understandable that the associated annotation can be displayed only to user a who initiated the content detection, or to some or all members of group 2, which is not limited in this application.

[0122] It can be seen that by associating and marking the summary of the detection and analysis report corresponding to the interactive message and displaying it on the corresponding interactive message, on the one hand, it is convenient for users to check the authenticity of the corresponding interactive message at any time in a targeted manner; on the other hand, it enables users to quickly understand the authenticity of the interactive message intuitively without the need for self-checking or relevant knowledge thresholds, which greatly improves the user experience and is especially suitable for the era of information explosion.

[0123] Optionally, in some embodiments of the present application, the content detection result of any candidate element seen by the user can be dynamically and intelligently detected and displayed directly based on the tracking detection of the user's gaze. Specifically, the method further includes:

[0124] Starting a camera in communication with the terminal to detect a first gaze area of ​​a user facing the first display interface;

[0125] If the first gaze area includes candidate elements that conform to a preset modality, selecting a target detection method adapted to the modality to perform content detection on at least one candidate element in the first gaze area according to the modality of each candidate element in the first gaze area;

[0126] If it is detected that the first gaze area includes a third target element that conforms to at least one preset generative artificial intelligence content generation method, a third annotation is set and displayed for the third target element, and the third annotation is used to display the confidence, generation method and data source of the third target element as generative artificial intelligence content.

[0127] like Figure 6 As shown, after user b sends the document "XX video processing method" and picture a in group 1, the eyes of user a correspond to the first gaze area in the mobile phone display interface, triggering the start of the content recognition program, and identifying the interactive message in the first gaze area as picture a. Then, the content detection result of picture a can be displayed in the conversation interface of user a's group 1, for example, marking the local or global area in picture a that meets AIGC.

[0128] It can be seen that by intelligently tracking the user's gaze and performing intelligent content detection on the candidate elements of any modality in the first gaze area corresponding to the user's target, the user does not need to trigger the content detection process through manual touch, voice interaction, etc., and the purpose of intelligent content detection can be achieved, greatly improving the user experience. The user can start or close this function at any time, and it can coexist with the aforementioned triggering of the content detection process through manual touch, voice interaction, etc., and there is no limit on the combination method.

[0129] Optionally, in some embodiments of the present application, the method can also be applied to virtual reality scenes, such as VR page flipping and interacting with objects in a virtual reality screen based on a handheld control terminal. The user wears a virtual reality helmet, which can be divided into a target tracking detection scene and a handheld controller detection scene. The details are as follows:

[0130] 1. Detection scenes based on gaze tracking

[0131] Specifically include:

[0132] The user's gaze is tracked and detected in real time, and a second gaze area corresponding to the user's gaze is dynamically locked; if it is detected that the second gaze area includes a fourth target element that conforms to at least one preset generative artificial intelligence content generation method, a fourth annotation is set and displayed for the third target element, and the fourth annotation is used to display the confidence, generation method and data source of the fourth target element as generative artificial intelligence content.

[0133] Among them, real-time tracking and detection of the user's gaze, that is, dynamically finding and tracking the focus of the eyeball, means that the camera detects the gaze area on the screen. If the area includes candidate objects that meet the preset conditions, the appropriate preset detection method is selected for these candidate objects for content detection.

[0134] 2. Detection scene based on handheld controller

[0135] The method is also used to:

[0136] The handheld control terminal of the virtual reality helmet is tracked and detected in real time, and the operation of the handheld control terminal on the virtual reality screen is dynamically detected; if the operation triggers the content detection function, the current virtual reality screen is detected; if the second gaze area includes a fourth target element that conforms to at least one preset generative artificial intelligence content generation method, a fourth annotation is set and displayed for the third target element, and the fourth annotation is used to display the confidence, generation method and data source of the fourth target element as generative artificial intelligence content.

[0137] It can be seen that in a virtual reality scene, when users are immersed in it and interact with various objects or materials, they can also understand the materials generated based on AIGC in the virtual reality screen as if they were in the scene, and can also objectively evaluate these materials generated based on AIGC.

[0138] Optionally, in some embodiments of the present application, the present solution may also be applied to adversarial training tasks. For example, the first target element may be a picture that meets preset business requirements. The method further includes:

[0139] Obtaining content description information of the first target element;

[0140] According to the content description information of the preset business requirement, obtaining a preset confrontation pattern matching the content description information from a preset confrontation pattern library;

[0141] The first target element and the preset adversarial pattern are synthesized, and at least one adversarial sample obtained by the synthesis is updated into an adversarial sample library, where the adversarial sample is used for adversarial training.

[0142] Specifically, the generated image is automatically synthesized with a preset adversarial pattern that matches the text (it can also be a trigger in text adversarial) to obtain an adversarial sample, and more adversarial samples can also be expanded for adversarial training. Specifically, after semantic analysis of the text, an adaptive adversarial pattern can be selected; or, element analysis of the generated image can be performed to extract image description information, and an adaptive adversarial pattern can be selected based on the image description information and business needs, and then synthesized so that the generated image is not identified as a target image after being crawled by an AI search engine (for example, for binding identity or clustering of certain types of images). It is especially suitable for retrieval systems. If the user chooses text recognition, it detects whether the text paragraph selected by the user is AIGC; if the user chooses image recognition, it detects whether the image selected by the user is AIGC; if the user chooses web page recognition, it detects whether the current web page is AIGC; AIGC paragraphs in the article can be annotated, or local areas (AIGC) in the image can be annotated.

[0143] It can be seen that in this way, pictures that are more suitable for adversarial training tasks can be collected anytime and anywhere, so as to generate and enrich, or update the adversarial sample library, so that the adversarial training can keep pace with the times, thereby further improving the security of the model.

[0144] For ease of understanding, take a specific application scenario (WeChat chat) as an example, refer to Figure 3-4d , specifically including the following operations:

[0145] The detection tool runs in the background of the mobile phone as a plug-in, or is inserted into a web page, and can be embedded in an input method or content recognition tool.

[0146] 1: The user browses any content in the WeChat group conversation and inputs a touch command for the content that the user wants to detect in the WeChat group conversation, that is, the user's touch command for the target element in the content is detected.

[0147] The arbitrary content may include: WeChat chat lists, mobile phone pictures, public account articles, web page articles, and content of any APP interface.

[0148] 2: Respond to the touch command and identify the target element according to the media type contained in the target element, such as text (segmentation), picture (or converting text into picture), and identify the target element that meets at least one preset AIGC generation method.

[0149] Specifically, after the user selects the target content, he / she can jump to the detection page of the detection tool to display a more comprehensive detection and analysis report. It is suitable for news and document retrieval systems on mobile phones (checking for duplicate content, checking for non-intellectual achievements, and other non-original situations).

[0150] Alternatively, in the small window that pops up for pattern recognition, click the icon "Detect AIGC content" to perform the detection. The detection results can be displayed in the small window and associated with the target element, or the small window can be hidden and the associated mark can be displayed directly in the area corresponding to the target element in the original text.

[0151] The same applies to the detection of images, audio, and video. You only need to select the appropriate preset detection method. I will not go into details here.

[0152] 3: Dynamically annotate the target element, which displays: the confidence, generation method, and source of the AIGC content.

[0153] Figures 1 to 6 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 7 to 9 The corresponding embodiments will not be described in detail later.

[0154] The above is a description of a method for processing interactive generative artificial intelligence content in an embodiment of the present application. The following is an introduction to a content detection device that executes the above-mentioned method for processing interactive generative artificial intelligence content.

[0155] See also Figure 7 ,like Figure 7 The structural diagram of a content detection device 40 shown in the figure can be applied to detect content forgery in offline media content of any mode, online media content of any mode, online voice conversations, online video conversations, interactive messages generated by group conversations of instant messaging applications, etc., and prompt the user. The content detection device 40 in the embodiment of the present application can realize the corresponding Figure 1-Figure 6 The steps in the interactive generative artificial intelligence content processing method performed by the content detection device 40 in any corresponding embodiment. The functions implemented by the 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 content detection device 40 may include an input-output module 401, a processing module 402, and a display module 403. The implementation of the functions of the input-output module 401, the processing module 402, and the display module 403 can be referred to Figure 1-Figure 6 The operations performed in any corresponding embodiment are not described in detail here.

[0156] In some embodiments, the input-output module 401 may be used to obtain a first touch input of a user to a first area of ​​a first display interface, where the first area includes at least one candidate element of at least one modality to be detected by the user;

[0157] The processing module 402 may be configured to scan the first area based on machine vision technology in response to the first touch input acquired by the input / output module 401, obtain at least one candidate element of at least one modality, and display the candidate element in a pop-up first window through the display module 403;

[0158] The input-output module 401 is further used to obtain a second touch input from a user for any candidate element in the first window;

[0159] The processing module 402 is further configured to, in response to the second touch input acquired by the input / output module 401, control the display module 403 to jump to the second display interface of the content detection tool, and select a target detection method adapted to the modality to perform content detection on at least one of the candidate elements according to the modality of each candidate element in the first window;

[0160] If it is detected that the first area includes a first target element that conforms to at least one preset generative artificial intelligence content generation method, a first annotation is set for the first target element and displayed through the display module 403, and the first annotation is used to display the confidence, generation method and data source of the first target element as generative artificial intelligence content.

[0161] Other functions can be found Figure 1-Figure 6 Any corresponding embodiments in the description will not be repeated here.

[0162] The above describes the content detection device 40 for executing the generative artificial intelligence content processing method in the embodiment of the present application from the perspective of modular functional entities. The following describes the content detection device 40 for executing the generative artificial intelligence content processing 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 content detection device 40 shown may have Figure 8 The structure shown, when Figure 7 The content detection device 40 shown has the following Figure 8 When the structure shown is Figure 8 The processor and transceiver in the content detection device 40 can realize the same or similar functions as the input / output module 401 and the processing module 402 provided in the above-mentioned device embodiment corresponding to the content detection device 40. Figure 8 The memory in the storage processor is a computer program that needs to be called when executing the above-mentioned generative artificial intelligence content processing method.

[0163] The present application also provides another content detection device, such as Fig. 9 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 content detection device can be any 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 content detection device (English full name: Point of Sales, English abbreviation: POS), a car computer, etc., taking the content detection device as a mobile phone as an example:

[0164] Fig. 9 FIG. 1 is a block diagram of a partial structure of a mobile phone related to the content detection device provided in an embodiment of the present application. Fig. 9 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 Fig. 9 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.

[0165] Combine the following Fig. 9 A detailed introduction to the various components of the mobile phone:

[0166] 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.

[0167] 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.

[0168] 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.

[0169] 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.

[0170] 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.

[0171] 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.

[0172] 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. Fig. 9 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.

[0173] 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.

[0174] 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.

[0175] Although not shown, the mobile phone may also include a camera, a Bluetooth module, etc., which will not be described in detail here.

[0176] 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 Fig. 9 The steps performed by the content detection device 40 in the above embodiment can be based on the mobile phone structure shown in Figure 109. For example, the processor 722 performs the following operations by calling the instructions in the memory 732:

[0177] Acquiring, by the input unit 730, a first touch input of a user on a first area of ​​a first display interface, where the first area includes at least one candidate element of at least one modality to be detected by the user;

[0178] In response to the first touch input acquired by the input unit 730, the first area is scanned based on machine vision technology to obtain at least one candidate element of at least one modality and display it in a pop-up first window through the display unit 740;

[0179] The input unit 730 is also used to obtain a second touch input from a user for any candidate element in the first window;

[0180] In response to the second touch input acquired by the input unit 730, the display unit 740 is controlled to jump to the second display interface of the content detection tool, and according to the modality of each candidate element in the first window, a target detection method adapted to the modality is selected to perform content detection on at least one of the candidate elements;

[0181] If it is detected that the first area includes a first target element that conforms to at least one preset generative artificial intelligence content generation method, a first annotation is set for the first target element and displayed through the display unit 740, and the first annotation is used to display the confidence, generation method and data source of the first target element as generative artificial intelligence content.

[0182] 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.

[0183] 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.

[0184] 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.

[0185] 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.

[0186] 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.

[0187] 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.

[0188] 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.

[0189] 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 method for processing interactive generative artificial intelligence content, characterized in that: The method is performed by a content detection tool, which runs in at least one of the following ways: in the terminal background, inserted into a web page, or embedded in an input method in the form of a plug-in; the method includes: Acquire a first touch input of a user on a first area of ​​a first display interface, where the first area includes at least one candidate element of at least one modality to be detected by the user; In response to the first touch input, scanning the first area based on machine vision technology to obtain at least one candidate element of at least one modality and display it in a pop-up first window; Acquire a second touch input of a user for any candidate element in the first window; In response to the second touch input, jump to the second display interface of the content detection tool, and select a target detection mode adapted to the modality to perform content detection on at least one of the candidate elements according to the modality of each candidate element in the first window; If it is detected that the first area includes a first target element that conforms to at least one preset generative artificial intelligence content generation method, a first annotation is set and displayed for the first target element, and the first annotation is used to display the confidence, generation method and data source that the first target element is generative artificial intelligence content.

2. The method according to claim 1, characterized in that: In response to the second touch input, jumping to the second display interface of the content detection tool, selecting a target detection method adapted to the modality to perform content detection on at least one of the candidate elements according to the modality of each candidate element in the first window; if it is detected that the first area includes a first target element that conforms to at least one preset generative artificial intelligence content generation method, setting a first annotation for the first target element and displaying it, including: In response to the second touch input, a new window for pattern recognition is popped up; In response to a user clicking an input on a generative artificial intelligence AIGC content detection function icon in the new window, performing content detection on the selected candidate element; Displaying the obtained detection result in the new window; The detection result is associated with a first target element that conforms to at least one preset generative artificial intelligence content generation method in a first annotation manner; or after hiding the new window, the detection result is associated and annotated with the first annotation in the area corresponding to the first target element in the original text to which the first target element belongs.

3. The method according to claim 1, characterized in that The method is applied to detect data of any modality appearing in any way in an instant messaging application; The first display interface is a session interface in which the user is currently participating, and the session interface is a single session interface or a group session interface; The method further comprises: Acquire historical conversation elements of any mode generated in the first display interface within a target time period, wherein the conversation elements include at least one of text, document, picture, video and voice; According to the modality of the historical conversation elements, selecting a target detection method adapted to the modality to perform content detection on at least one of the historical conversation elements of the at least one modality; If a second target element that conforms to at least one preset generative artificial intelligence content generation method is detected in the historical conversation element, a second annotation is set for the second target element, and the second annotation is used to display the confidence, generation method and data source of the second target element as generative artificial intelligence content.

4. The method according to claim 3, characterized in that The step of selecting a target detection method adapted to the modality to perform content detection on at least one historical conversation element of the at least one modality according to the modality of the historical conversation element includes: After determining that there is an interactive message matching a preset keyword in the conversation interface within the target time period, obtaining a candidate element of any modality matching the preset keyword in the interactive message, where the interactive message is a document, a picture, or a video; According to the mode of the interactive message, the content of each candidate element is detected; The detection results are annotated and displayed on the corresponding interactive message.

5. The method according to claim 1, 3 or 4, characterized in that: The method is applied to a news and document retrieval system displayed in real time on a terminal, and the method further comprises: Detecting a third touch input of the user on the conversation interface to the target interactive message; In response to the third touch input, displaying a content detection window of the target interactive message; In response to the user selecting the target content in the content detection window, jumping to the detection page of the content detection tool and displaying the corresponding detection analysis report; The summary of the detection and analysis report is associated and annotated and displayed on the corresponding interactive message.

6. The method according to claim 1, characterized in that The first target element is a picture that meets a preset business requirement, and the method further includes: Obtaining content description information of the first target element; According to the content description information of the preset business requirement, obtaining a preset confrontation pattern matching the content description information from a preset confrontation pattern library; The first target element and the preset adversarial pattern are synthesized, and at least one adversarial sample obtained by the synthesis is updated into an adversarial sample library, where the adversarial sample is used for adversarial training.

7. The method according to claim 1, characterized in that The method further comprises: Starting a camera in communication with the terminal to detect a first gaze area of ​​a user facing the first display interface; If the first gaze area includes candidate elements that conform to a preset modality, selecting a target detection method adapted to the modality to perform content detection on at least one candidate element in the first gaze area according to the modality of each candidate element in the first gaze area; If it is detected that the first gaze area includes a third target element that conforms to at least one preset generative artificial intelligence content generation method, a third annotation is set and displayed for the third target element, and the third annotation is used to display the confidence, generation method and data source of the third target element as generative artificial intelligence content.

8. The method according to claim 1, characterized in that The method is applied to a virtual reality scene, the user wears a virtual reality helmet, and the method further includes: Track and detect the user's gaze in real time, and dynamically lock the second gaze area corresponding to the user's gaze; If it is detected that the second gaze area includes a fourth target element that conforms to at least one preset generative artificial intelligence content generation method, a fourth annotation is set and displayed for the third target element, and the fourth annotation is used to display the confidence, generation method and data source of the fourth target element as generative artificial intelligence content.

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 8.

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 8.

Images