Decentralized multimodal deep-FAKE detection

Abstract

An electronic device for decentralized multimodal deepfake detection, is disclosed. The electronic device receives media content and information including content type or complexity. The electronic device determines a content-based identifier for the media content and stores the content-based identifier on a distributed ledger. The electronic device applies executable code on the information based on the stored identifier. The electronic device selects, from a set of deep learning models, a deep learning model based on the applied code. The electronic device generate a verification result including a deep-fake indicator, based on application of the selected deep learning mode. The electronic device receives first content from a trusted-content database based on the verification result. The electronic device stores the verification result on the distributed ledger based on the content-based identifier and the first content. The electronic device controls a display device to render the deep-fake indicator, based on the stored verification result.

Description

DECENTRALIZED MULTIMODAL DEEP-FAKE DETECTIONCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to Indian Provisional Application No. IN202411051085, which was filed on July 03, 2024. The above stated Patent Application(s) are hereby incorporated herein by reference in their entirety.FIELD OF INVENTION

[0002] Various embodiments of the disclosure relate to decentralized multimodal deepfake detection systems. More specifically, various embodiments of the disclosure relate to an electronic device and a method for detecting deepfakes across various media types using blockchain technology and artificial intelligence.BACKGROUND

[0003] The rapid advancement of digital technology and artificial intelligence has led to the emergence of sophisticated content manipulation techniques, particularly in the realm of deepfake creation. Deepfakes, which involve the use of machine learning methods to generate or manipulate audio-visual content, have become increasingly prevalent across various media platforms. This technology has found applications in entertainment, education, and creative industries, but has also raised concerns regarding the potential for misinformation and fraud. Conventional approaches to detecting manipulated content often rely on centralized systems and single-model detection methods. The conventional approaches typically involve analysis of visual or auditory artifacts, inconsistencies in metadata, or patterns indicative of artificial generation. However, such approaches face limitations in terms of scalability, adaptability to new manipulation techniques, and the ability to handle diverse types of media content. Additionally, centralized detection systems may be vulnerable to single point of failure andlack transparency in their decision-making processes, that potentially undermines trust in the verification results.

[0004] Further limitations and disadvantages of conventional and traditional approaches will become apparent to one of skill in the art, through comparison of described systems with some aspects of the present disclosure, as set forth in the remainder of the present application and with reference to the drawings.SUMMARY

[0005] An electronic device and method for decentralized multimodal deep-fake detection, is provided substantially as shown in, and / or described in connection with, at least one of the figures, as set forth more completely in the claims.

[0006] These and other features and advantages of the present disclosure may be appreciated from a review of the following detailed description of the present disclosure, along with the accompanying figures in which like reference numerals refer to like parts throughout.BRIEF DESCRIPTION OF FIGURES

[0007] FIG. 1 is a block diagram of a network environment for decentralized multimodal deep-fake detection, in accordance with an embodiment of the disclosure.

[0008] FIG. 2 is a block diagram that illustrates an exemplary electronic device of FIG.1 , in accordance with an embodiment of the disclosure.

[0009] FIG. 3A and FIG. 3B collectively illustrate an exemplary architecture for a system for decentralized multimodal deep-fake detection, in accordance with one embodiment of the disclosure.

[0010] FIG. 4 is a sequence diagram of a process for detection and verification of media content, in accordance with another embodiment of the disclosure.

[0011] FIG. 5 is a diagram that illustrates a first exemplary scenario for content analysis, in accordance with another embodiment of the disclosure.

[0012] FIG. 6 is a diagram that illustrates a second exemplary scenario for content analysis, in accordance with at least one embodiment of the disclosure.

[0013] FIG. 7 is a diagram that illustrates an example scenario of a VeriGuard uploader interface, in at least one embodiment of the disclosure.

[0014] FIG. 8 is a diagram that illustrates an exemplary scenario for a media analysis interface, in accordance with another embodiment of the disclosure.

[0015] FIG. 9 is a flowchart that illustrates operations of an exemplary method for decentralized multimodal deep-fake detection, in accordance with an embodiment of the disclosure.DETAILED DESCRIPTION

[0016] The following described implementation may be found in an electronic device and a method for decentralized multimodal deepfake detection. Exemplary aspects of the disclosure may provide an electronic device, which may include a circuitry that may be configured to receive media content and first information including at least one of a content type or a content complexity associated with the media content. The circuitry may further be configured to determine a content-based identifier associated with the media content. The circuitry may further be configured to store the content-based identifier on a distributed ledger. The circuitry may further be configured to apply an executable code associated with the distributed ledger on the first information, based on the stored content-based identifier. The circuitry may further be configured to select, from a set of deep learning models, a first deep learning model based on the applied executable code. The first deep learning model may be configured to generate a verification result including a deep-fake indicator of the media content. The circuitry may further be configured to receive, from a trusted-content database, first content associated with the media content, based on the verification result. The circuitry may further be configured to store the verification result on the distributed ledger, based on the content-based identifier and the first content. Thecircuitry may further be configured to control rendering of the deep-fake indicator of the media content to at least one of a user or an owner associated with the media content, based on the stored verification result.

[0017] Conventional approaches to the detection of manipulated content rely on centralized systems and single-model detection methods, which have limitations in terms of scalability, adaptability to new manipulation techniques, and the ability to handle diverse types of media content. These centralized detection systems may be vulnerable to single point of failure and lack transparency in their decision-making processes, that potentially undermines trust in the verification results. As the sophistication of deepfake technology continues to advance, there is a critical need for a more robust, adaptable, and transparent solution that may effectively detect and verify manipulated content across various media types.

[0018] The disclosed electronic device may leverage blockchain technology and artificial intelligence to provide a comprehensive and transparent approach to content verification. Unlike traditional centralized systems, the disclosed method may utilize a distributed ledger to store content identifiers and verification results, to ensure immutability and transparency of the detection process. The disclosed electronic device may employ a dynamic selection of deep learning models based on content type and complexity, to allow for more accurate and adaptable detection across various media formats. The dynamic selection of deep learning models may result in enhanced scalability through decentralized processing, improved trust and transparency through blockchain integration, and increased accuracy in detecting diverse types of manipulated content through the use of multiple specialized deep learning models.

[0019] FIG. 1 is a block diagram of a network environment for decentralized multimodal deep-fake detection, in accordance with an embodiment of the disclosure. With reference to FIG. 1 , there is shown an exemplary network environment 100. The networkenvironment 100 includes an electronic device 102, a server 104, a media database 106A, a trusted-content database 106B, a communication network 108, media content 110, a distributed ledger 112, a set of deep learning models 116, and a user device 118. The set of deep learning models 116 may include a deep learning model 116A, a deep learning model 116B, ... and a deep learning model 116N. The distributed ledger 112 may include executable codes, such as, an executable code 114A, an executable code 114B, ... and an executable code 114N.

[0020] It should be noted that “N” number of deep learning models and executable code shown in FIG. 1 are for exemplary purposes, and the scope of the disclosure should not be so limited. The set of deep learning models 116 may include only 2 or more than “N” deep learning models, and similarly the distributed ledger 112 may include only 2 or more than “N” executable code, without departure from the scope of the disclosure.

[0021] The electronic device 102 may include suitable logic, circuitry, interfaces, and / or code that may be configured to receive media content 110 and first information including at least one of a content type or a content complexity associated with the media content 110. The media content 110 may include text content, audio content, image content, video content, augmented reality / virtual reality (AR / VR) content, multi-media content, and the like. The electronic device 102 may further be configured to determine a content-based identifier associated with the media content 110. The electronic device 102 may further be configured to store the content-based identifier on the distributed ledger 112. The electronic device 102 may further be configured to apply an executable code associated with the distributed ledger 112 on the first information, based on the stored content-based identifier. The electronic device 102 may further be configured to select from the set of deep learning models 116, a first deep learning model (for instance, the deep learning model 116A) based on the applied executable code. The electronic device 102 may further be configured to apply the selected deep learning model 116A on the media content 110.The electronic device 102 may further be configured to generate a verification result including a deep-fake indicator of the media content 110, based on the application of the selected deep learning model 116A. The electronic device 102 may further be configured to receive, from a trusted-content database (e.g., the trusted-content database 106B), first content associated with the media content 110, based on the verification result. The electronic device 102 may further be configured to store the verification result on the distributed ledger 112, based on the content-based identifier and the first content. The electronic device 102 may further be configured to control a display device (for instance, the user device 118), associated with at least one of a user or an owner associated with the media content, to render the deep-fake indicator of the media content 110, based on the stored verification result.

[0022] The electronic device 102 may be connected to the server 104, the media database 106A, the trusted-content database 106B, and the distributed ledger 112, through the communication network 108. Examples of the electronic device 102 may include, but are not limited to, a digital media player (DMP), a micro-console, a TV tuner, a digital media streamer, a media extender / regulator, a smart TV, a gaming console, a digital media hub, a computer workstation, a mainframe computer, a handheld computer, a smart phone, a mobile phone, a tablet computer, a personal computer, a smart appliance, a plug-in device, and / or any other computing device with content streaming functionality.

[0023] The electronic device 102 may store the set of deep learning models 116 or may be remotely connected to another system (such as the server 104) that hosts the set of deep learning models 116. When hosted on another system, the electronic device 102 may send instructions to control training or inference of the set of deep learning models 116 via remote calls (e.g., application programming interface (API) calls).

[0024] The server 104 may include suitable logic, circuitry, interfaces, and / or code that may be configured to host the media database 106A. In some scenarios, the server 104 may also host the trusted-content database 106B. In addition, the server may be associated with the distributed ledger 112. In an embodiment, the server 104 may be configured to detect that the media content 110 is deep-fake content based on a multimodal decentralized technique. The server 104 may execute the operations of the electronic device 102 for the decentralized deep-fake content detection.

[0025] The server 104 may be implemented as a cloud server and may execute operations through web applications, cloud applications, HTTP requests, repository operations, file transfer, and the like. Other example implementations of the server 104 may include, but are not limited to, a database server, a file server, a web server, a media server, an application server, a mainframe server, a machine learning server (enabled with or hosting, for example, a computing resource, a memory resource, and a networking resource), or a cloud computing server.

[0026] In at least one embodiment, the server 104 may be implemented as a plurality of distributed cloud-based resources by use of several technologies that are well known to those ordinarily skilled in the art. A person with ordinary skill in the art will understand that the scope of the disclosure may not be limited to the implementation of the server 104 and the electronic device 102, as two separate entities. In certain embodiments, the functionalities of the server 104 can be incorporated in its entirety or at least partially in the electronic device 102 without a departure from the scope of the disclosure. In certain embodiments, the server 104 may host the media database 106A. Alternatively, the server 104 may be separate from the media database 106A and may be communicatively coupled to the media database 106A. In other embodiments, the server 104 may also be communicatively coupled to the trusted-content database 106B. Alternatively, the server 104 may be separate host the trusted-content database 106B.

[0027] The media database 106A may include suitable logic, interfaces, and / or code that may be configured to store path or address of the media content 110. The media content 110 may include at least one of text content, audio content, image content, video content, augmented reality / virtual reality (AR / VR) content, or multi-media content. The media database 106A may be derived from data off a relational or non-relational database, or a set of comma-separated values (csv) files in conventional or big-data storage. The media database 106A may be stored or cached on a device, such as a server (e.g., the server 104) or the electronic device 102. The device storing the media database 106A may be configured to receive commands or instructions from the electronic device 102 or the server 104. In response, the device storing the media database 106A may be configured to retrieve and provide the first information associated with the media content 110.

[0028] In some embodiments, the media database 106A may be hosted on a plurality of servers stored at the same or different locations. The operations of the media database 106A may be executed using hardware including a processor, a microprocessor (e.g., to perform or control performance of one or more operations), a field-programmable gate array (FPGA), or an application-specific integrated circuit (ASIC). In some other instances, the media database 106A may be implemented using software.

[0029] The trusted-content database 106B may include suitable logic, interfaces, and / or code that may be configured to store first content associated with the media content 110. The trusted-content database 106B may also store second content associated with the media content 110. The trusted-content database 106B may include a personal insights module (e.g., personal insights module 322 of FIG. 3B). The first content and the second content may include image content, text content, or audio content, and may pertain to trusted content associated with the media content 110.The trusted content may refer to verified, credible, authenticated, and reliable content. The trusted-content database 106B may be derived from data off a relational or non-relational database, or a set of comma-separated values (csv) files in conventional or big-data storage. The trusted-content database 106B may be stored or cached on a device, such as a server (e.g., the server 104) or the electronic device 102. The device storing the trusted-content database 106B may be configured to receive commands or instructions from the electronic device 102 or the server 104. In response, the device of the trusted-content database 106B may be configured to retrieve and provide the authenticated content.

[0030] In some embodiments, the trusted-content database 106B may be hosted on a plurality of servers stored at the same or different locations. The operations of the trusted- content database 106B may be executed using hardware including a processor, a microprocessor (e.g., to perform or control performance of one or more operations), a field- programmable gate array (FPGA), or an application-specific integrated circuit (ASIC). In some other instances, the trusted-content database 106B may be implemented using software.

[0031] The communication network 108 may include a communication medium through which the electronic device 102 and the server 104 (and / or the media database 106A, the trusted-content database 106B, and the distributed ledger 112) may communicate with one another. The communication network 108 may be one of a wired connection or a wireless connection. Examples of the communication network 108 may include, but are not limited to, the Internet, a cloud network, Cellular or Wireless Mobile Network (such as Long-Term Evolution and 5thGeneration (5G) New Radio (NR)), satellite communication system (using, for example, low earth orbit satellites), a Wireless Fidelity (Wi-Fi) network, a Personal Area Network (PAN), a Local Area Network (LAN), or a Metropolitan Area Network (MAN). Various devices in the network environment 100 may be configured to connect to the communication network 108 in accordance with various wired and wireless communication protocols. Examples of such wired and wireless communication protocols may include, but are not limited to, at least one of a Transmission Control Protocol andInternet Protocol (TIP / IP), User Datagram Protocol (UDP), Hypertext Transfer Protocol (HTTP), File Transfer Protocol (FTP), Zig Bee, EDGE, IEEE 802.11 , light fidelity (Li-Fi), 802.16, IEEE 802.11 s, IEEE 802.11 g, multi-hop communication, wireless access point (AP), device to device communication, cellular communication protocols, and Bluetooth (BT) communication protocols.

[0032] The distributed ledger 112 may be a decentralized and distributed database system that may maintain an immutable record of data operations or transactions. The distributed ledger 112 may include a network of nodes 112A where each node of the network of nodes 112A may include a laptop, a smartphone, a mainframe computer, and other such mobile computing devices. A set of data operations may be grouped together as a block and may be further linked to a previous block of data operations to form a chain of a plurality of blocks. All blocks of data operations may be stored in a decentralized manner, whereby at least two participants or nodes of the distributed ledger 112 may store a sub-set of a plurality of blocks associated with one or more transactions in which the at least two participants or nodes may participate. Each node of the network of nodes 112A may independently verify the transaction’s validity, ensuring that the transaction meets the necessary criteria and that there are no conflicts or duplications. After validation, the transaction may be grouped with other transactions into a block, which is then added to the existing chain of blocks on the distributed ledger 112. In an instance, the block associated with the transaction may be added to the existing chain of blocks through a consensus mechanism, such as proof of work or proof of stake. Further, the distributed ledger 112 may include an operating system which may allow for deployment of a smart contract between multiple parties, for example, a user interface (for instance, a user interface 306 of FIG. 3) of a user (for instance, a user 304 of FIG. 3) and the electronic device 102.

[0033] The distributed ledger 112 may be a chain of blocks which mat use accounts as state objects and a state of each account may be tracked by the chain. Herein, the accounts represent identities of users, mining nodes, or automated agents. All the blocks of data operations or the smart contract are associated with the accounts on the chain of blocks. By way of example, and not limitation, the distributed ledger 112 may be an Ethereum blockchain which may use accounts as state objects and a state of each account can be tracked by the Ethereum blockchain. Herein, the accounts represent identities of users, mining nodes, or automated agents. All the blocks of data operations or the smart contract are associated with the accounts on the Ethereum Blockchain. The scope of the disclosure may not be limited to the implementation of the distributed ledger 112 as the Ethereum blockchain, a Hyperledger blockchain, or a Corda blockchain. Other implementations of the distributed ledger 112 may be possible in the present disclosure, without a deviation from the scope of the present disclosure.

[0034] The distributed ledger 112 may serve as a decentralized database for storing content-based identifiers and verification results, associated with the media content 110. The distributed ledger 112 may contain multiple executable codes, including the executable code 114A, the executable code 114B, ... and the executable code 114N. An executable code (for instance, the executable code 114A) may be applied on the first information, based on the stored content-based identifier. Each executable code of the executable codes may interact with the communication network 108 to process and verify the media content 110. In some embodiments, the distributed ledger 112 may be implemented using blockchain technology, where each block may contain a cryptographic hash of the previous block, a timestamp, and transaction data. By way of example, and not limitation, the distributed ledger 112 may be a Corda blockchain, an Ethereum blockchain, or a Hyperledger blockchain. In some embodiments, the distributed ledger 112 may utilize alternative distributed ledger technologies such as directed acyclic graphs(DAGs) or hybrid systems combining multiple ledger types. The executable codes (114A, 114B ... 114N) associated with the distributed ledger 112 may be implemented as smart contracts, which are self-executing codes that may automatically enforce and execute the terms of an agreement between parties. The smart contract may include a set of conditions under which the parties to the smart contract may agree to interact with each other. The smart contract may run on one or more nodes of the distributed ledger 112 and may govern transitions.

[0035] The set of deep learning models 116 may include a plurality of deep learning models 116A, 116B ... 116N, where each deep learning model of the set of deep learning models 116 may be configured to analyze a distinct content type associated with the media content 110, such as, text content, image content, video content, ARA / R content, or multimedia content. A deep learning model (for instance, the deep learning model 116A) may be selected from the set of deep learning models 116 based on the applied executable code (for instance, executable code 114A). Further, the selected deep learning model 116A may be applied to the media content 110, such that the circuitry 202, based on the application of the deep learning model 116A on the media content 110, may analyze the media content 110 to generate a verification result. In an instance, the verification result may include a deep-fake indicator of the media content 110. Further, the generated verification result may be stored in the distributed ledger 112. The deep fake indicator may include signs or clues that may help identify whether a video, audio, or image has been manipulated using deep learning techniques to create realistic but deep fake content. The detection of deep fake content may involve analysis of various aspects of the media content 110, such as inconsistencies in visual or audio elements, unnatural movements, anomalies in the background or lighting, and the like. In an instance, presence of unnatural facial movements or expressions may be a deep fake indicator. In another instance, if a video shows a person speaking but their lip movements do not perfectly synchronize withcorresponding audio, or facial expressions of the person may seem exaggerated or robotic, which may be a deep fake indicator and act as a sign of deep fake manipulation. In another instance, inconsistencies in lighting or shadows on a person’s face, which may not match the rest of the scene, may be a deep fake indicator. The inconsistencies may occur because deep fake operations may not perfectly replicate the complex interplay of light and shadow in a real environment.

[0036] Another example scenario may involve audio deep fakes, where a person’s voice may be synthesized to say things they never actually said. The deep fake indicator of audio deep fakes may include unnatural intonations, awkward pauses, or inconsistencies in the background noise. For instance, if a recording of a speech has sudden changes in the ambient sound or the speaker’s voice lacks the natural variations in pitch and tone, it may be a deep fake indicator of audio manipulation. By being aware of these indicators, individuals and organizations can better protect themselves from the potential harm caused by deep fakes.

[0037] Though the above embodiments elaborate application of a single deep learning model, however it should be noted that more than one deep learning model may be applied at a time on the media content 110, without departure from the scope of the disclosure.

[0038] Each deep learning model of the set of deep learning models 116 may be a hybrid network, which may include multiple neural networks. Further, each of the multiple neural networks may be a computational network or a system of artificial neurons, arranged in a plurality of layers, as nodes. The plurality of layers of the neural network may include an input layer, one or more hidden layers, and an output layer. Each layer of the plurality of layers may include one or more nodes (or artificial neurons). Outputs of all nodes in the input layer may be coupled to at least one node of hidden layer(s). Similarly, inputs of each hidden layer may be coupled to outputs of at least one node in other layers of the neural network. Outputs of each hidden layer may be coupled to inputs of at leastone node in other layers of the neural network. Node(s) in the final layer may receive inputs from at least one hidden layer to output a result. The number of layers and the number of nodes in each layer may be determined from hyper-parameters of the neural network. Such hyper-parameters may be set before or after training the neural network on the training dataset.

[0039] Each neural network of the neural networks may include electronic data, which may be implemented as, for example, a software component of an application executable on the electronic device 102. Each of the neural networks may rely on libraries, external scripts, or other logic / instructions for execution by a processing device, such as the electronic device 102. Further, each of the neural networks may rely on code and routines to enable a computing device, such as the electronic device 102 to perform one or more operations, such as deep-fake detection in the media content 110. In some embodiments, each of the neural networks may be implemented using hardware including a processor, a microprocessor (e.g., to perform or control performance of one or more operations), a field-programmable gate array (FPGA), or an application-specific integrated circuit (ASIC). Alternatively, in some embodiments, each of the neural networks may be implemented using a combination of hardware and software.

[0040] In some embodiments, the set of deep learning models 116 may include generative adversarial networks (GANs) to detect synthetic content in the media content 110, or attention-based models to identify inconsistencies in case the media content 110 is complex. The selection and application of the deep learning models (116A, 116B ... 116N) may be dynamically adjusted based on a type of the media content 110 and information associated with content complexity of the media content 110.

[0041] Examples of each of the set of deep learning models 116 may include, but are not limited to, convolutional neural networks, recurrent neural networks, transformer models, generative adversarial networks, and ensemble models. For example, theconvolutional neural networks may be used for image-based deepfake detection, and analysis of spatial features and patterns in visual content. Recurrent neural networks may be utilized for analysis of sequential data, such as audio or video frames, and capture of temporal dependencies in the media content. Transformer models may be applied for both image and text analysis, based on self-attention mechanisms to identify inconsistencies or artifacts indicative of synthetic content generation. Generative adversarial networks, while primarily used for content generation, may also be adapted for deepfake detection by training a discriminator model to distinguish between real and synthetic content. Ensemble models may combine multiple neural network architectures, each of which may specialize in different aspects of deepfake detection, to improve overall accuracy and robustness.

[0042] The user device 118 may include a user-interface through which a user may interact with the electronic device 102, feed commands and instructions, and receive deepfake indicator from the electronic device 102. The user device 118 may be fixed at a place or may be portable. Examples of the user device 118 may include, but are not limited to, a smartphone, a touchpad, a personal computer, a wearable device, an infotainment system, an in-vehicle display, and a voice-controlled device.

[0043] In operation, the electronic device 102 may be configured to receive the media content 110 and first information including at least one of a content type or a content complexity associated with the media content 110. The media content 110 may include at least one of text content, audio content, image content, video content, augmented reality / virtual reality (AR / VR) content, multi-media content, and the like. The content complexity depends on the content type and a number of the content type in the media content 110.

[0044] Further, the first information may include the content type of the media content110. In another embodiment, the first information may include content complexityassociated with the media content 110. In another embodiment, the first information may include both the content type and the content complexity associated with the media content 110. In an instance, the content complexity associated with the media content 110 may be defined in terms of various levels of complexity, for instance, low, medium, and high. Content complexity may be determined by evaluation of quantitative measures like readability scores, content type, size of content, and the like. In an instance, the content complexity in text content may be determined based on length of content, word count, and sentence length. In an embodiment, low complexity content may feature simple content type, small size, and well-marked structures. Medium complexity content may involve more intricate content type and structures and may be of medium size. High complexity content may include complex and implicit content type and structures of significant size.

[0045] In an example, the electronic device 102 may receive an article (not shown) that may include text only as the media content 110. For the received article, the first information may include content type of the article as the text content. Further, the first information may include content complexity associated with the article as low.

[0046] In another example, the electronic device 102 may receive a paper column (not shown) that may include text along with an image as the media content 110 (as shown in FIG. 5A). For the received paper column, the first information may include content type of the article as a combination of the text content and the image content. Further, the first information may include content complexity associated with the article as medium.

[0047] In another example, the electronic device 102 may receive a clip (not shown) that may include an augmented reality (AR) based video as the media content 110. For the received clip, the first information may include the content type of the article as a combination of the AR content and the video content. Further, the first information may include content complexity associated with the clip as high.

[0048] The electronic device 102 may receive the media content 110 through a userinterface. Alternatively, the electronic device 102 may communicate with the server 104 and receive the media content 110 through the server 104. Alternatively, the electronic device 102 may retrieve the media content 110 from the media database 106A. Alternatively, the media content 110 may be transmitted to the electronic device 102 through a remote or external server.

[0049] The electronic device 102 may be configured to determine a content-based identifier associated with the media content 110. The electronic device 102 may analyze intrinsic properties of the media content 110 and the first information to generate a unique content-based identifier (for instance, a hash value) that may be used to index, retrieve, and track the media content 110. The process of analysis may typically include examination of content type of the media content 110, such as text content, audio content, video content, or image content. In an instance, for text content, the process of analysis may involve keyword extraction, semantic analysis, or natural language processing to identify the intrinsic properties of the media content 110, for instance, unique phrases or topics. In another instance, in case of video content, the process of analysis may include detection of the intrinsic properties of the media content 110, for instance, sound patterns, speech recognition, or visual elements like color histograms, shapes, and motion vectors. In another instance, in case of image content, the process of analysis may include detection of the intrinsic properties of the media content 110, for instance, pattern recognition, edge detection, texture analysis, and the like.

[0050] Further, based on the analysis, the electronic device 102 may generate a hash value of the media content 110,

[0051] Hash value of the media content 110 may be generated by applying a mathematical function, known as a hash function, to the media content 110. The electronic device 102, through the hash function, may process the media content 110 and convert into a fixed-length string of characters, regardless of original data size of the media content110. The process of generation of the hash value may involve dividing data associated with the media content 110 into equal-sized blocks and running the hash function on each block, often multiple times, to produce a unique hash value. Further, the electronic device 102 may determine the content-based identifier based on the generated hash value. In an instance, the hash value may be combined with additional metadata, such as codec used to encode the data and the hash function, to determine the content-based identifier. The process of determination of the content-based identifier ensures that any change in the media content 110 may result in a completely different content-based identifier, which leads to a reliable gateway for verifying data integrity and authenticity. For example, in a decentralized storage, such as Interplanetary File System (for instance, Interplanetary File System (IPFS) 312 in FIG. 3), the content-based identifier may be used to uniquely identify and retrieve the media content 110 based on corresponding hash value, ensuring that the exact version of the media content 110 may be accessed. The content-based identifier may serve as a digital fingerprint for the media content 110 and may enable efficient organization, search, and retrieval of the media content 110 in large databases.

[0052] The electronic device 102 may be further configured to store the content-based identifier on the distributed ledger 112. The electronic device 102 may further store the media content 110 on a decentralized storage, for instance, the media storage 312. The electronic device 102 may further associate the media content 110 with the content-based identifier stored on the distributed ledger 112. To store the content-based identifier on the distributed ledger 112, the electronic device 102 may prepare the content-based identifier for storage based on a format or data structure that may be efficiently recorded on the distributed ledger 112. The distributed ledger 112 may provide a decentralized and immutable record-keeping system, to ensure that once the content-based identifier is stored, the content-based identifier may not be altered or tampered with.

[0053] Based on a format of the content-based identifier, the electronic device 102 mayadd the formatted content-based identifier to a transaction that may be broadcasted to the network of nodes 112A of the distributed ledger 112. After validation, the transaction may be grouped with other transactions into a block, which is then added to the existing chain of blocks on the distributed ledger 112. The process ensures that the content-based identifier may be securely stored across the network of nodes 112A, that may provide redundancy and resilience against data loss or corruption. Further, distributed nature of the distributed ledger 112 may also facilitate easy access and retrieval of the contentbased identifier, enabling efficient content management and verification across various applications and platforms.

[0054] In some embodiments, the electronic device 102 may further receive second information, which may include a source of the media content 110, a timestamp associated with the media content 110, or metadata associated with the media content 110. The electronic device 102 may receive the second information from the server 104 or the media database 106A. Further, the electronic device 102 may store the second information on the distributed ledger 112 and associate the second information with the content-based identifier.

[0055] The electronic device 102 may further be configured to apply an executable code associated with the distributed ledger 112 on the first information. The distributed ledger 112 may contain multiple executable codes, including the executable code 114A, the executable code 114B, ... and the executable code 114N that may be executed on the media content 110 to execute precise operations based on the specific media content 110. The electronic device 102 may select an executable code (for instance, executable code 114A) from the multiple executable codes. The selection of the executable code 114A may be based on content type or content complexity. The executable code 114A may be applied on the first information, based on the stored content-based identifier. The application of the executable code 114A may ensure that the operations performed by theelectronic device 102 are precise and based on the specific media content 110.

[0056] The electronic device 102 may be further configured to select from the set of deep learning models 116, a first deep learning model based on the applied executable code. The set of deep learning models 116 may include a plurality of deep learning models 116A, 116B ... 116N, where each deep learning model of the plurality of deep learning models 116A, 116B ... 116N may be configured to analyze different content type of the media content 110. In an instance, the electronic device 102 may apply the deep learning model 116A and may process text content based on the application of the deep learning model 116A. In another instance, the electronic device 102 may apply the deep learning model 116B and may process audio content based on the application of the deep learning model 116B. In yet another instance, the electronic device 102 may apply the deep learning model 116N and may process multimedia content based on the application of the deep learning model 116N.

[0057] The electronic device 102 may select a second deep learning model (for instance, the deep learning model 116B) from the set of deep learning models 116 based on the applied executable code (for instance, the executable code 114B). Further, the electronic device 102 may generate the verification result based on outputs of the deep learning model 116A and the deep learning model 116B.

[0058] In some embodiments, for the media content 110 including multiple content types, an array of executable codes (114A ... 114N) may be applied on the first information. Further, the electronic device 102 may select multiple deep learning models from the set of deep learning models 116, based on the applied executable codes. In an instance, in case the media content 110 includes audio content and multimedia content, the deep learning models 116B and 116N may be selected from the set of deep learning models 116. Further, the electronic device 102 may apply the selected deep learning models (116B and 116N) to process corresponding content of the media content 110, andmay correspondingly generate a verification result, which may include a deep-fake indicator of the media content 110. The selected deep learning models (116B and 116N) may be applied simultaneously with one another, in cascade, or in series (one-by-one or one-after-another) with one another based on requirement.

[0059] In an instance, for the media content 110 including images, the selected deep learning model may be EfficientNet, which may deliver high accuracy while maintaining computational efficiency, making it ideal for deepfake detection at scale. The EfficientNet may use significantly fewer parameters than traditional deep learning models. The EfficientNet may efficiently balance depth, width, and resolution, reducing computational overhead while achieving results comparable to or better than heavyweight models like ResNet and DenseNet. The EfficientNet may generalize well, avoiding overfitting and maintaining high performance across diverse deepfake manipulation techniques.

[0060] In another instance, for the media content 110 including videos, the selected deep learning model may be lnceptionResNetV2, which may handle high computational demand of video analysis while ensuring efficiency, accuracy, and scalability. The lnceptionResNetV2 may capture both spatial and temporal inconsistencies in deepfake videos, which helps in detecting manipulations across frames.

[0061] In another instance, for the media content 110 including text, the selected deep learning model may be a dynamic text detection module, which may address the nature of language models, need for adaptability, and importance of explainability in classification decisions. The text detection module may include a neural network classifier, which in turn, may include a combination of transformer-based language models (BERT, GPT) with linguistic feature analysis to detect patterns beyond probability scores, improving decision reliability. The text detection module may further include a linguistic feature extractor, which may analyze sentence structure, lexical diversity, punctuation usage, and formality. The linguistic feature extractor may also differentiate between Al-generated and humantext based on writing characteristics, he text detection module may further include a RADAR Adversarial module to generate adversarial Al-generated text, which helps to detect sophisticated Al writing tactics. The text detection module may further include FewShot Adapter, which enables the text detection module to learn from minimal examples of new Al-generated texts, adapt rapidly to new language models, and reduce the need for extensive retraining, improving efficiency and scalability.

[0062] Further, the electronic device 102 may determine a confidence score associated with the verification result. In an instance, a high confidence score (for instance, confidence score more than 90%) may indicate that the verification result is dependable. In another instance, a low confidence score (for instance, confidence score less than 90%) may indicate that the verification result is unreliable. In such case, associated deep learning models are required to retrained or updated.

[0063] The verification result may determine whether the media content 110 in question is genuine or has been tampered with. Further, the deep-fake indicator may be a marker or signal that identifies whether the media content 110 or a part of the media content 110 has been manipulated using deep-fake technology. By including the deep-fake indicator in the verification result, the electronic device 102 may assess the authenticity of the media content 110 or at least a part of the media content 110. In an embodiment, in case the indicator suggests that the media content 110 is a deep-fake, the electronic device 102 may execute appropriate actions, such as flagging the media content 110 for further review or rejecting media content 110 altogether. Hence, the deep-fake indicator ensures that only verified and trustworthy media content or a part of the media content 110 is retrieved and used, thereby maintaining the integrity and reliability.

[0064] The electronic device 102 may be further configured to receive, from the trusted- content database 106B, first content associated with the media content 110, based on the verification result. The electronic device 102 may further receive, from the trusted-contentdatabase 106B, second content associated with the media content 110. The trusted- content database may act as a secure and reliable source that may store various content types associated with the media content 110, that may ensure authenticity and integrity of the content. The electronic device 102 may communicate with the trusted-content database 106B and may request and receive the first and second content associated with the particular part of the media content 110 based on the verification result. The first content and the second content may be at least one of text content, audio content, image content, video content, augmented reality / virtual reality (AR / VR) content, or multi-media content. In an embodiment, the electronic device 102 may compare the first content and the second content. Further, the electronic device 102 may update the verification result based on the comparison. In an instance, the first content and the second content may represent specific portions of the media content 110. For instance, the first content may pertain to text content of the media content 110, and the second content may pertain to image content of the media content 110. Further, the electronic device 102 may compare the first content and the second content to relate the first content with the second content, i.e. , to determine significance or consistency of the text content in combination with the image content. Further, the electronic device 102 may update the verification result based on the determined significance or consistency of the text content in combination with the image content.

[0065] In another embodiment, the electronic device 102 may receive an update to the set of deep learning models 116. In an embodiment, the electronic device 102 may receive input from the user device 118 associated with at least one of the users, developers, or owners . The electronic device 102 may further generate update based on the received input . Further, the electronic device 102 may re-select a deep learning model based on the update to the set of deep learning models 116. Furthermore, the electronic device 102 may apply the re-selected deep learning model on the media content 110 and may updatethe verification result based on the application of the re-selected deep learning model.

[0066] In an instance, the electronic device 102 may generate a notification including the verification result. The electronic device 102 may further transmit the notification to a second electronic device (for instance the user device 118) associated with the at least one of the user, the developer, or the owner.

[0067] The electronic device 102 may be configured to store the verification result on the distributed ledger 112, based on the content-based identifier and the first content. The electronic device 102 may further store a confidence score on the distributed ledger 112 and associate the confidence score with the verification result. In an instance, the verification result may act as a block, which may be further linked to a previous block of data operations to form a chain of a plurality of blocks associated with the distributed ledger 112. Further, all the blocks of data operations may be stored on the distributed ledger 112 in a decentralized manner. The electronic device 102 may further generate a distributed ledger transaction including the verification result and may broadcast the distributed ledger transaction to the network of nodes 112A associated with the distributed ledger 112.

[0068] The electronic device 102 may be configured to control the user device 118 associated with at least one of a user or an owner associated with the media content 110, to render the deep-fake indicator of the media content 110. In an instance, the electronic device 102 may receive a request from a user to access the media content 110. The electronic device 102 may further verify the request based on the stored verification result. In an embodiment, the electronic device 102 may not verify the request in case the verification result have a low confidence score. In another embodiment, the electronic device 102 may verify the request in case the verification result has high confidence score.

[0069] The electronic device 102 may further control access to the media content based on the verification of the request. The electronic device 102 may transmit information associated with the deep-fake indicator of the media content 110 to the user device 118(of the user or the owner) linked with the electronic device 102 The user devices may include a smartphone, a touchpad, a GUI interface, a personal computer, a microphone, or a display device associated with the user or the owner associated with the media content 110. The electronic device 102 may receive a user input based on the rendered deep-fake indicator. The electronic device 102 may further update the verification result stored on the distributed ledger 112 based on the received user input.

[0070] Unlike conventional techniques that rely on centralized systems for content verification and deep-fake detection, the disclosed electronic device 102 leverages the distributed ledger 112 and the set of deep learning models 116 for secure and efficient analysis of the media content 110. Further, the disclosed electronic device 102 offers advantages such as improved adaptability to various content types, enhanced transparency through blockchain integration, and increased accuracy in detection of manipulated content across diverse media formats.

[0071] FIG. 2 is a block diagram that illustrates an exemplary electronic device of FIG.1 , in accordance with an embodiment of the disclosure. FIG. 2 is described in conjunction with elements from FIG. 1. With reference to FIG. 2, there is shown an exemplary block diagram 200 of the electronic device 102. The electronic device 102 may include circuitry 202, a memory 204, a network interface 206, and an input / output (I / O) device 208. The I / O device 208 may include a display device 208A. The memory 204 may include the media content 110. The network interface 206 may connect the electronic device 102 with the server 104, via the communication network 108.

[0072] The circuitry 202 may include suitable logic, circuitry, and / or interfaces that may be configured to execute program instructions associated with different operations to be executed by the electronic device 102. The operations may include, for instance, media content reception, content-based identifier determination, content-based identifier storage, executable code application, deep learning model selection, first content reception,verification result storage, deep-fake indicator rendering, and the like. The circuitry 202 may include one or more processing units, which may be implemented as a separate processor. In an embodiment, the one or more processing units may be implemented as an integrated processor or a cluster of processors that perform the functions of the one or more specialized processing units, collectively. The circuitry 202 may be implemented based on a number of processor technologies known in the art. Examples of implementations of the circuitry 202 may be an X86-based processor, a Graphics Processing Unit (GPU), a Reduced Instruction Set Computing (RISC) processor, an Application-Specific Integrated Circuit (ASIC) processor, a Complex Instruction Set Computing (CISC) processor, a microcontroller, a central processing unit (CPU), and / or a combination thereof.

[0073] The memory 204 may include suitable logic, circuitry, interfaces, and / or code that may be configured to store one or more instructions to be executed by the circuitry 202. The one or more instructions stored in the memory 204 may be executed to perform the different operations of the circuitry 202 (and / or the electronic device 102). The memory 204 may be further configured to store the media content 110. Examples of implementation of the memory 204 may include, but are not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Hard Disk Drive (HDD), a Solid-State Drive (SSD), a CPU cache, and / or a Secure Digital (SD) card.

[0074] The network interface 206 may include suitable logic, circuitry, interfaces, and / or code that may be configured to facilitate communication between the electronic device 102 and the server 104 (and / or the media database 106A, the trusted-content database 106B, and the distributed ledger 112), via the communication network 108. The network interface 206 may be implemented by use of various known technologies to support wired or wireless communication of the electronic device 102 with the communication network108. The network interface 206 may include, but is not limited to, an antenna, a radio frequency (RF) transceiver, one or more amplifiers, a tuner, one or more oscillators, a digital signal processor, a coder-decoder (CODEC) chipset, a subscriber identity module (SIM) card, or a local buffer circuitry.

[0075] The network interface 206 may be configured to communicate via wireless communication with networks, such as the Internet, an Intranet, a wireless network, a cellular telephone network, a wireless local area network (LAN), or a metropolitan area network (MAN). The wireless communication may be configured to use one or more of a plurality of communication standards, protocols and technologies, such as Global System for Mobile Communications (GSM), Enhanced Data GSM Environment (EDGE), wideband code division multiple access (W-CDMA), Long Term Evolution (LTE), 5thGeneration (5G) New Radio (NR), code division multiple access (CDMA), time division multiple access (TDMA), Bluetooth, Wireless Fidelity (Wi-Fi) (such as IEEE 802.11 a, IEEE 802.11 b, IEEE 802.11 g or IEEE 802.11 n), voice over Internet Protocol (VoIP), light fidelity (Li-Fi), Worldwide Interoperability for Microwave Access (Wi-MAX), a protocol for email, instant messaging, and a Short Message Service (SMS).

[0076] The I / O device 208 may include suitable logic, circuitry, interfaces, and / or code that may be configured to receive an input from a user and provide an output to the user based on the received input. For example, the I / O device 208 may receive the media content 110 and first information including at least one of a content type or a content complexity associated with the media content 110. The I / O device 208 may be further configured to render the deep-fake indicator of the media content 110 on the user interface, for instance, a user device. Examples of the I / O device 208 may include, but are not limited to, a display (e.g., a touch screen), a keyboard, a mouse, a joystick, a microphone, or a speaker. Examples of the I / O device 208 may further include braille I / O devices, such as, braille keyboards and braille readers.

[0077] The display device 208A may include suitable logic, circuitry, and interfaces that may be configured to display or render the deep-fake indicator of the media content 110. In some embodiments, the display device 208A may be a touch screen which may enable a user to provide a user-input via the display device 208A. The display device 208A may be realized through several known technologies such as, but not limited to, at least one of a Liquid Crystal Display (LCD) display, a Light Emitting Diode (LED) display, a plasma display, or an Organic LED (OLED) display technology, or other display devices. In accordance with an embodiment, the display device 208A may refer to a display screen of a head mounted device (HMD), a smart-glass device, a see-through display, a projectionbased display, an electro-chromic display, or a transparent display. Various operations of the circuitry 202 are described further, for example, in FIG. 3.

[0078] FIG. 3A and FIG. 3B collectively illustrate an exemplary architecture for a system for decentralized multimodal deep-fake detection, in accordance with one embodiment of the disclosure. FIG. 3A and FIG. 3B are described in conjunction with elements from FIG. 1 and FIG. 2. With reference to FIG. 3A and FIG. 3B, there is shown an architecture 300 for a system for decentralized multimodal deep-fake detection. The architecture 300 includes input options 302, a user interface 306, a blockchain network 308, a backend service 310, a media storage 312, a deepfake detection module 314, an intelligent media extractor 316, an operation selection module 318, a set of deep learning models 320, personal insights module 322, processing module 324, content owner notifier 326, and memory 328. The input options 302 may include various input options, for instance, an image input option, a video with audio input option, a video without audio input option, a text input option, an image with text input option, a text with image input option, and the like. The user interface 306 may provide the input options 302. The user interface 306 may facilitate upload and detection of the media content 110 based on interaction with a user (for instance, user 304). The user interface 306 may be implemented as part of the I / Odevice 208 of the electronic device 102. FIG. 3 also shows a user 304 who may be associated with and operate the electronic device 102, through the user interface 306.

[0079] The circuitry 202 may receive, via the input options 302, the media content 110. The circuitry 202 may receive, via the image input option, image content. Alternatively, the circuitry 202 may receive, via the video with audio input option, video with audio content. Alternatively, the circuitry 202 may receive, via the text input option, text content. In some embodiments, the input options 302 may also accommodate other types of media content 110, such as audio content or augmented reality / virtual reality (ARA / R) content.

[0080] The user interface 306 may be configured to interface with the user 304 and enable selection of input types from the input options 302. In an instance, the user interface 306 may provide a single button click mechanism for the user 304 to upload media and initiate detection. The user interface 306 may further display information such as upload status, blockchain transactions (e.g., CID logging), detection results (e.g., whether the content is a deepfake), and notifications about duplicate files or existing results. The user interface 306 may communicate with the blockchain network 308 and the backend service 310. In an instance, the user device 118 may act as the user interface 306.

[0081] The blockchain network 308 may be an exemplary implementation of the distributed ledger 112. The blockchain network 308 may store content-based identifiers associated with the media content 110. In some embodiments, the blockchain network 308 may interact with the backend service 310 to trigger detection processes and log detection results(for instance, deep-fake indicator, verification result, and confidence score) associated with the media content 110. In an instance, the blockchain network 308 may log the content-based identifiers generated by the media storage 312 and corresponding detection results as immutable records. The blockchain network 308 may further signal the backend service 310 to initiate the detection process when a new content-based identifier is logged. The blockchain network 308 may also implementsolidity-based contracts to log new uploads and detection results and facilitate retrieval of stored data (e.g., uploader wallet address, prior detection results, and the like).

[0082] The backend service 310 may be configured to activate deepfake detection operations through the operation selection module 318. The backend service 310 may apply an executable code (for instance, the executable code 114A of FIG. 1 ) associated with the distributed ledger 112 on the information received with the media content 110, based on the stored content-based identifier. The backend service 310 may transmit detection results (e.g., verified content or deepfake) to a frontend service, and further the frontend service may transmit the detection results to the user interface 306 for display.

[0083] The media storage 312 may be implemented as a decentralized storage system. In some embodiments, the media storage 312 may use Interplanetary File System (IPFS) technology. The IPFS technology may facilitate storage of files associated with the uploaded media content 110 securely and may generate a unique Content Identifier (CID) for each file based on its content. The IPFS technology may further ensure that duplicate files (with identical content but different names) are not uploaded multiple times by verifying if a CID already exists. The media storage 312 may store the media content 110 in association with the content-based identifier stored on the blockchain network 308. The media storage 312 may generate and store media content identifiers (CIDs) for the uploaded media content 110.

[0084] The deepfake detection module 314 may include the intelligent media extractor 316, the operation selection module 318, and the set of deep learning models 320. The intelligent media extractor 316 may implement an intelligent classification mechanism that dynamically distinguishes between simple and complex content types, including embedded or multi-modal content. The intelligent media extractor 316 may automatically identify major features like text from images, facial features from videos, and audio from videos and other multi modal formats. The intelligent media extractor 316 may further filterout information and focus only contextually significant features based on the content type and ensure that the set of deep learning models 320 receive optimal features as input for prediction.

[0085] The operation selection module 318 may implement a dynamic operation selection process that adapts based on the content type of media content 110 and content complexity. The operation selection module 318 may leverage deep learning strategies to autonomously select an optimal model of the set of deep learning models 320 based on the media content 110, content type, and content complexity, ensuring precision in analysis and detection. In an exemplary embodiment, the operation selection module 318 may select an optimal deep learning model from the set of deep learning models 320 based on the applied executable code 114A. in an instance, the set of deep learning models 320 may correspond to the set of deep learning models 116. The electronic device 102 may apply the selected deep learning model to generate a verification result including a deep-fake indicator associated with the media content 110. In an instance, the operation selection module 318 may identify that the uploaded media content 110 contains multiple content types (e.g., text embedded within images or images inside a PDF) and may automatically select an appropriate operation to process each content type. The operation selection module 318 may also cross-check the detection results with entries in the blockchain network 308. The operation selection module 318 may further provide verification metadata if the media content 110 is already uploaded.

[0086] The deepfake detection module 314 may constitute a part of the circuitry 202 that may facilitate seamless execution of the deployed set of deep learning models 320, maintaining computational efficiency while processing diverse media content 110.

[0087] The personal insights module 322 may store the detection results in a firebase database to keep track of the media content 110 that has been uploaded. The personal insights module 322 may be a separate module that provides personal insights like datauploaded and notifications for the user 304. The personal insights module 322 may be a part of the circuitry 202. The processing module 324 may be a sub-component of the personal insights module 322, which may process the uploaded media content 110 and aid in maintaining the track.

[0088] The content owner notifier 326 may be an exemplary implementation of the user device 118. The content owner notifier 326 may be configured to display upload status of the media content 110, the deep-fake indicator, and the verification result to notify content owners or users. In some embodiments, the content owner notifier 326 may render the deep-fake indicator associated with the media content 110 to at least one of a user or an owner associated with the media content 110, based on the stored verification result.

[0089] The memory 328 may store data history including references to previously analyzed media content 110 and associated detection results. The memory 328 may aid in display of the uploaded media content 110 personally for each user if the media content 110 has already been analyzed and logged in the content owner notifier 326.

[0090] The components of the architecture 300 may interact to process and verify media content 110 in the following manner. The user interface 306 may upload the media content 110 to the circuitry 202 through the input options 302 based on commands of the user 304. The media storage (i.e., the IPFS media storage 312) may store the media content 110 and may generate a content-based identifier. The blockchain network 308 may store the generated content-based identifier and initiate an operation of the backend service 310. The backend service 310 may apply executable code 114A associated with the distributed ledger 112 on the information (i.e., the first information) received with the media content 110. The deepfake detection module 314 of the circuitry 202 may identify major features of the media content 110 through the intelligent media extractor 316. The deepfake detection module 314 may activate the operation selection module 318 based on the application of the executable code 114A. The circuitry 202, through the operationselection module 318, may select an appropriate deep learning model of the set of deep learning models 320based on the type and complexity of the media content 110. The selected deep learning model may generate a verification result including a deep-fake indicator. The electronic device 102 may compare the verification result with content from the personal insights module 322. The blockchain network 308 may store the verification result, based on the content-based identifier and the first content from the personal insights module 322. The personal insights module 322 may store the detection results to keep track of the media content 110 that has been uploaded. The circuitry 202, through the content owner notifier 326, may render the deep-fake indicator to the user 304 or content owner based on the stored verification result. Thus, the decentralized architecture 300 may provide a robust and transparent technique to detect and verify digital content across various types of the media content 110.

[0091] FIG. 4 is a sequence diagram of a process for detection and verification of media content, in accordance with another embodiment of the disclosure. FIG. 4 is described in conjunction with elements from FIG. 1 , FIG. 2, and FIG. 3. With reference to FIG. 4, there is shown an exemplary sequence diagram 400. The sequence diagram 400 includes a user interface 402, input options 404, a media storage 406, a deepfake detection module 408, a blockchain network 410, a backend service 412, a smart decision pipeline module 414, a dynamic operation selection module 416, a set of deep learning models 418, personal insights module 420, a content owner notifier 422, and data history memory 424. The deepfake detection module 408, the backend service 412, the smart decision pipeline module 414, the dynamic operation selection module 416, the personal insights module 420, the content owner notifier 422, and the data history memory 424 may be components of the circuitry 202.

[0092] The sequence diagram 400 illustrates the interactions between various components of the electronic device 102 during the process of media content verification.The process begins when the user interface 402 interacts with the circuitry 202 to upload the media content 110 on the circuitry 202 based on commands given by user. The user interface 402 may be implemented as part of the I / O device 208 of the electronic device 102.

[0093] The user interface 402 may provide the input options 404, which may include various input options, for instance, an image input option, a video input option, a text input option, and the like.

[0094] The media storage 406 may store the media content 110. The media storage 406 may be an exemplary implementation of the media storage 312, as described in FIG.3. In an instance, the media storage 406 may correspond to IPFS storage. In some embodiments, the media storage 406 may generate a hash value of the media content 110. The circuitry 202 may be configured to determine a content-based identifier based on the generated hash value.

[0095] After the storage of the media content 110, the deepfake detection module 408 may process the media content 110 for analysis. The deepfake detection module 408 may be an implementation of the deepfake detection module 314.

[0096] Further, the blockchain network 410 may store content identifier associated with the media content stored at the media storage 406. The circuitry 202 may log transactions, requests, and results at the blockchain network 410 . The blockchain network 410 may be implemented as the distributed ledger 112.

[0097] In some embodiments, the circuitry 202 may be configured to receive second information including at least one of a source of the media content 110, a timestamp associated with the media content 110, or metadata associated with the media content 110. The circuitry 202 may be further configured to store the second information on the blockchain network 410. The circuitry 202 may further associate the stored second information with the content-based identifier.

[0098] The blockchain network 410 may be an exemplary implementation of the distributed ledger 112. The blockchain network 410 may store content-based identifiers associated with the media content 110. then trigger the backend service 412 to detect processes and log detection results (for instance, deep-fake indicator and verification result) associated with the media content 110. The backend service 412 may correspond to the backend service 310, as described in FIG. 3. .

[0099] The smart decision pipeline module 414 may select an algorithm for smart decision based on interaction with the backend service 412. The selected algorithm may further activate the deepfake detection module 408. The deepfake detection module 408 may select a deep learning model from the set of deep learning models 418. The deepfake detection module 408 may select a set of operations through the dynamic operation selection module 416. The dynamic operation selection module 416 may implement a dynamic operation selection process that adapts based on the content type of media content 110 and content complexity. The functionality of the dynamic operation selection module 416 may be similar to the operation selection module 318.

[0100] In some embodiments, the deepfake detection module 408 may be configured to select a first deep learning model and a second deep learning model from the set of deep learning models 418 based on the applied executable code. The dynamic operation selection module 416 may further activate the selected deep learning model(s). The deepfake detection module 408 may then process the media content 110 using the selected deep learning models.

[0101] The circuitry 202 may generate a verification result (referred to as analysis results, herein), which may include a deep-fake indicator for the media content 110. In some embodiments, the circuitry 202 may be configured to generate the verification result based on the application of the first deep learning model and the second deep learning model. In an instance, the circuitry 202 may be further configured to determine aconfidence score associated with the verification result. This confidence score may be stored on the backend service 412 in association with the verification result.

[0102] The backend service 412 may then update the blockchain network 410 with the verification result. The circuitry 202 may store the verification result on the blockchain network 410. The circuitry 202 may further associate the verification result with the content-based identifier of the media content 110.

[0103] The personal insights module 420 may sync user data from the deepfake detection module 408. The personal insights module 420 may be a separate module that provides personal insights like data uploaded and notifications for the user. The functioning of the personal insights module 420 may be similar to that of the personal insights module 322.

[0104] The circuitry 202, through the deepfake detection module 408, may report potential deepfake content, which may be notified at the content owner notifier 422 . The circuitry 202, through the content owner notifier 422, may access the personal insights module 420 to retrieve the verification data for comparison. Further, the personal insights module 420 may notify an upload or reupload of the media content 110, through the content owner notifier 422.

[0105] In some embodiments, the circuitry 202 may be configured to control the user device 118 to render the deep-fake indicator of the media content based on the stored verification result. The rendering may be done through the user interface 402. The deepfake indicator may be rendered to report the media content 110 as a potential deep fake. The content owner notifier 422 may notify the upload or reupload at the user interface 402.

[0106] The data history memory 424 may store personal history of the user by retrieving the personal history from the personal insights module 420. Further, the personal history of the user may be retrieved from the data history memory 424 and may be displayed at the user interface 402, based on request.

[0107] The sequence diagram 400 illustrates a comprehensive process for media content verification, which may leverage Blockchain technology for secure storage of verification results and multiple deep learning models for accurate deepfake detection across various types of the media content 110.

[0108] In some embodiments, the circuitry 202 may be configured to receive second information including at least one of a source of the media content 110, a timestamp associated with the media content 110, or metadata associated with the media content 110. The circuitry 202 may be further configured to store the second information on the distributed ledger 112 in association with the content-based identifier.

[0109] In some embodiments, the circuitry 202 may be configured to receive, from the trusted-content database 106B, second content associated with the media content 110. The circuitry 202 may be further configured to compare the first content and the second content. Based on this comparison, the circuitry 202 may be configured to update the verification result.

[0110] In some embodiments, the circuitry 202 may be configured to store the confidence score on the distributed ledger 112. The circuitry 202 may further associate the confidence score with the verification result. The circuitry 202 may be further configured to control the user device 118 to render the confidence score associated with the verification result to at least one of the user 304 or an owner associated with the media content 110.

[0111] In some embodiments, the circuitry 202 may be configured to receive a user input based on the rendered deep-fake indicator. The circuitry 202 may be further configured to update the verification result stored on the distributed ledger 112. The update of the verification result may be based on the received user input.

[0112] In some embodiments, the circuitry 202 may be configured to receive a request to access the media content 110. The circuitry 202 may be further configured to verify therequest based on the stored verification result. Based on this verification, the circuitry 202 may be configured to control access to the media content 110.

[0113] In some embodiments, the circuitry 202 may be configured to generate a distributed ledger transaction including the verification result. The circuitry 202 may be further configured to broadcast the distributed ledger transaction to the network of nodes 112A associated with the distributed ledger 112.

[0114] FIG. 5 is a diagram that illustrates a first exemplary scenario for content analysis, in accordance with another embodiment of the disclosure. FIG. 5 is described in conjunction with elements from FIG. 1 , FIG. 2, FIG. 3, and FIG. 4. With reference to FIG. 5, there is shown an exemplary scenario 500. The scenario 500 includes an input content frame 502, an extracted face image 504, a deepfake image detection module 506, extracted text content 508, a deepfake text detection module 510, and a display interface 512. The deepfake image detection module 506 and the deepfake text detection module 510 may be a part of the circuitry 202.

[0115] The input content frame 502 may contain text content 502-1 and image content 502-2. In some embodiments, the input content frame 502 may be received through the network interface 206 of the electronic device 102. The text content 502-1 and the image content 502-2 may be components of the media content 110.

[0116] The circuitry 202 may be configured to process the input content frame 502 to extract the face image 504 and the text content 508. The extracted face image 504 may be directed to the deepfake image detection module 506 for analysis. The deepfake image detection module 506 may be implemented using one or more deep learning models from the set of deep learning models 116. In some embodiments, the circuitry 202, through the deepfake image detection module 506, may be configured to detect artificially generated or manipulated image content.

[0117] The extracted text content 508 may be directed to the deepfake text detection module 510 for analysis. The deepfake text detection module 510 may also be implemented using one or more deep learning models from the set of deep learning models 116. In some embodiments, the circuitry 202, through the deepfake text detection module 510, may detect artificially generated or manipulated text content. In some embodiments, the circuitry 202, through the deepfake text detection module 510, may be configured to detect text embedded within images, which may allow for comprehensive analysis of complex media content.

[0118] The electronic device 102 may be configured to detect complex images containing both faces and text. In such embodiments, the circuitry 202, through the deepfake image detection module 506 and the deepfake text detection module 510, may analyze the respective components of the input content frame 502, based on the detected complex images.

[0119] The analyzed components of the input content frame 502 may be provided to the display interface 512. The display interface 512 may be implemented as part of the display device 208A. The circuitry 202 may control the display interface 512 to render the analyzed components, including any detected indicators of deepfake content in either the face image 504 or text content 508 of the input content frame 502.

[0120] The scenario 500 illustrates a parallel processing structure where both image content 502-2 and text content 502-1 may be analyzed simultaneously through their respective detection modules. The approach may allow for efficient and comprehensive analysis of complex media content, including cases where text may be embedded within images or where documents contain both textual and visual elements. It may be noted that the scenario 500 of FIG. 5 is for exemplary purposes and may not be construed to limit the scope of the disclosure.

[0121] FIG. 6 is a diagram that illustrates a second exemplary scenario for content analysis, in accordance with at least one embodiment of the disclosure. FIG. 6 is described in conjunction with elements from FIG. 1 , FIG. 2, FIG. 3, FIG. 4, and FIG. 5. With reference to FIG. 6, there is shown an exemplary second scenario 600. The second scenario 600 includes a content block 602, image content 604, a deepfake image detection module 606, text content 608, a deepfake text detection module 610, and a display interface 612. The deepfake image detection module 606 and the deepfake text detection module 610 may form a part of the circuitry 202.

[0122] The content block 602 may contain both the image content 604 and the text content 608 of the media content 110. In some embodiments, the circuitry 202 may receive the content block 602through the network interface 206 of the electronic device 102. The image content 604 and the text content 608 may be components of the media content 110. For example, the content block 602 may be extracted from a word processing file or a desktop publishing file. The electronic device 102 may be configured to detect images and text from these file formats and allow comprehensive analysis of various document types.

[0123] The circuitry 202 may be configured to process the content block 602 to separate the image content 604 and the text content 608 for individual analysis. The image content 604 may be directed to the deepfake image detection module 606 for analysis. The deepfake image detection module 606 may be implemented using one or more deep learning models from the set of deep learning models 116. In some embodiments, the deepfake image detection module 606 may detect artificially generated or manipulated image content.

[0124] The text content 608 may be directed to the deepfake text detection module610 for analysis. The deepfake text detection module 610 may also be implemented using one or more deep learning models from the set of deep learning models 116. In someembodiments, the deepfake text detection module 610 may detect artificially generated or manipulated text content.

[0125] The electronic device 102 may be configured to process mixed media content containing both images and text. In such embodiments, the circuitry 202 may activate both the deepfake image detection module 606 and the deepfake text detection module 610 to analyze the respective components of the content block 602.

[0126] The analyzed components of the content block 602may be combined and presented through the display interface 612. The display interface 612 may be implemented as part of the display device 208A. The display interface 612 may present the analyzed components, including any detected indicators of deepfake content in either the image content 604 or text content 608 of the content block 602.

[0127] In some embodiments, the circuitry 202 may be configured to receive, from the trusted-content database 106B, second content associated with the media content 110. The circuitry 202 may be further configured to compare the first content and the second content. Based on this comparison, the circuitry 202 may update the verification result.

[0128] The circuitry 202 may control the display interface 612 to render the updated verification result to the user 304. In some embodiments, the circuitry 202 may be configured to receive a user input based on the control of the rendered deep-fake indicator. The circuitry 202 may be further configured to update the verification result stored on the distributed ledger 112 based on the received user input.

[0129] The second scenario 600 illustrates a comprehensive approach to deepfake detection, where both image and text content may be analyzed simultaneously through their respective detection modules. This approach may allow for efficient and thorough analysis of complex media content, including cases where content may contain both visual and textual elements that require verification. It may be noted that the second scenario600 of FIG. 6 is for exemplary purposes and may not be construed to limit the scope of the disclosure.

[0130] FIG. 7 is a diagram that illustrates an example scenario of a VeriGuard uploader interface, in at least one embodiment of the disclosure. FIG. 7 is described in conjunction with elements from FIG. 1 , FIG. 2, FIG. 3, FIG. 4, FIG. 5, and FIG. 6. With reference to FIG. 7, there is shown an exemplary scenario 700. The scenario 700 represents an upload interface 702. The upload interface 702 may be configured to enable users to upload the media content 110 to the media storage 312, which may utilize Interplanetary File System (IPFS) technology. In an instance, the upload interface 702 may be a VeriGuard upload interface.

[0131] The upload interface 702 may include a wallet connection feature, which may be represented by a "Connect Wallet" button 702A. This feature may allow users to securely connect their digital wallets to the electronic device 102, which may provide identity verification and access control. In some embodiments, the wallet connection feature may display a dialog window 702B, showing connection options such as "Account A" and "Account B" for wallet selection.

[0132] The upload interface 702 may also include a media upload section 702C. The media upload section 702C may display a connection status, including a wallet address of the connected user. The media upload section 702C may provide a drag-and-drop area for image or video content selection, allowing users to easily upload the media content 110 to the electronic device 102.

[0133] In some embodiments, the circuitry 202 may be configured to generate a notification including the verification result after the media content 110 has been processed and analyzed by the set of deep learning models 116. The notification may include information such as the deep-fake indicator and any associated confidence scores.

[0134] The circuitry 202 may be further configured to transmit the notification to a second electronic device (for instance the user device 118) associated with the at least one of the users (for instance, user 304) or the owner of the media content 110.

[0135] The upload interface 702 may interact with other components of the electronic device 102, such as the blockchain network 308 and the backend service 310. When the media content 110 gets uploaded to the circuitry 202 through the upload interface 702, the media content 110 may be stored in the media storage 312, and a content-based identifier may be generated and stored on the distributed ledger 112.

[0136] In some embodiments, the circuitry 202, through the upload interface 702, may display confirmation messages showing successful media upload, including information such as the Content ID (CID) generated by the media storage 312 and the Transaction Hash stored on the blockchain network 308.

[0137] The upload interface 702 may provide a streamlined process for users to connect their digital wallets, upload the media content 110, and initiate the verification process for the decentralized multimodal deepfake detection. The upload interface 702 may enhance user experience and facilitate efficient interaction with the various components of the electronic device 102 for media content verification. It may be noted that the scenario 700 of FIG. 7 is for exemplary purposes and may not be construed to limit the scope of the disclosure.

[0138] FIG. 8 is a diagram that illustrates an exemplary scenario of a media analysis interface, in accordance with an embodiment of the disclosure. FIG. 8 is described in conjunction with elements from FIG. 1 , FIG. 2, FIG. 3, FIG. 4, FIG. 5, FIG. 6, and FIG. 7. With reference to FIG. 8, there is shown an exemplary scenario 800. The scenario 800 includes a media analysis interface 802 on the upload interface 702. The media analysis interface 802 may be configured to enable users to upload and analyze media contentthrough the electronic device 102. The media analysis interface 802 may be implemented as part of the display device 208A.

[0139] The media analysis interface 802 may further include an upload media section 802A where users may drag and drop or select media files for analysis. In some embodiments, the media analysis interface 802 may display a connection status 802B with an identifier, such as "0xe67f486f", which may represent a connected wallet address.

[0140] The media analysis interface 802 may include an "Upload and Analyze Media" button 802C. When the button (i.e. , the button 802C) is pressed, the circuitry 202 may be configured to receive the media content 110 and initiate the analysis process. The media analysis interface 802 may further display confirmation messages to show successful media upload, including information such as the Content ID (CID) generated by the media storage 312 and the Transaction Hash stored on the blockchain network 308.

[0141] In some embodiments, the circuitry 202 may be configured to receive a request to access the media content 110. The circuitry 202 may be further configured to verify the request based on the stored verification result. Based on the verification, the circuitry 202 may control access to the media content 110. In an instance, when the verification is successful, the circuitry 202 may allow access to the media content 110. In an instance, when the verification is unsuccessful, the circuitry 202 may deny access to the media content 110.The media analysis interface 802 may present a combined analysis result section 804, including both a deepfake (image or video) analysis component 806 and a text analysis component 808. In an instance, the combined analysis result section 804 may show two visualization components: an original media display 806A on the left side and a visualization 806B on the right side.

[0142] In some embodiments, the circuitry 202 may be configured to receive an update to the set of deep learning models 116. The circuitry 202 may be further configured to reselect a deep learning model from the updated set of deep learning models 116. Thecircuitry 202 may then update the verification result based on the application of the reselected deep learning model.

[0143] The media analysis interface 802 may indicate that the analyzed media is original or manipulated, along with a confidence score (as shown in the deepfake (image or video) analysis component 806). For example, the media analysis interface 802 may display that the analyzed media is likely original with a specific confidence percentage. Additionally, the text analysis component 808 may indicate whether the text in the media is human-written or artificially generated with a confidence score. The text analysis component 808 may further include extracted text 808A, for instance, “AN INTRIGUING FILM, MAGNIFICENT, ADVENTUROUS HILLS, UNMISSABLE, TENSE AND THRILLING”. The media analysis interface 802 and the text analysis component 808 may be a part of the circuitry 202.

[0144] The media analysis interface 802 may provide a clear visual representation of the analysis results, which may allow users to view both the original content and the analysis visualization side by side, along with confidence scores for both the media authenticity and text analysis. This comprehensive display may enable users to make informed decisions about the authenticity of the analyzed media content 110. It may be noted that the scenario 800 of FIG. 8 is for exemplary purposes and may not be construed to limit the scope of the disclosure.

[0145] FIG. 9 is a flowchart that illustrates operations of an exemplary method for decentralized multimodal deep-fake detection, in accordance with an embodiment of the disclosure. FIG. 9 is described in conjunction with elements from FIG. 1 , FIG. 2, FIG. 3, FIG. 4, FIG. 5, FIG. 6, FIG. 7, and FIG. 8. With reference to FIG. 9, there is shown an exemplary flowchart 900 for decentralized multimodal deep-fake detection. The flowchart 900 may include operations from 902 to 922 and may be implemented by the electronicdevice 102 of FIG. 1 or the circuitry 202 of FIG. 2. The flowchart 900 may start at 902 and proceed to 904.

[0146] At 904, media content 110 and first information may be received. The first information may include at least one of a content type or a content complexity associated with the media content 110. In some embodiments, the circuitry 202 may be configured to receive the media content 110 and first information through the network interface 206. Details related to the reception of media content 110 are described further, for example, in FIG. 3 and FIG. 4.

[0147] At 906, a content-based identifier associated with the media content 110 may be determined. In some embodiments, the circuitry 202 may be configured to generate a hash value of the media content 110 to determine the content-based identifier. Details related to the determination of the content-based identifier are described further, for example, in FIG. 3 and FIG. 4.

[0148] At 908, the content-based identifier may be stored on the distributed ledger 112. In some embodiments, the circuitry 202 may be configured to store the content-based identifier on the blockchain network 308. Details related to the storage of the contentbased identifier are described further, for example, in FIG. 3 and FIG. 4.

[0149] At 910, an executable code associated with the distributed ledger 112 may be applied on the first information, based on the stored content-based identifier. In some embodiments, the circuitry 202 may be configured to apply the executable code 114A, the executable code 114B, ... or the executable code 114N on the first information. Details related to the application of executable code are described further, for example, in FIG. 3.

[0150] At 912, a first deep learning model may be selected from the set of deep learning models 116 based on the applied executable code. In some embodiments, the circuitry 202 may be configured to select the deep learning model 116A, the deep learning model 116B, or the deep learning model 116N based on the applied executable code.Details related to the selection of deep learning models are described further, for example, in FIG. 3.

[0151] At 914, the selected first deep learning model may be applied on the media content 110. In an instance, the circuitry 202 may apply the deep learning model 116A and may process text content based on the application of the deep learning model 116A. In another instance, the circuitry 202 may apply the deep learning model 116B and may process audio content based on the application of the deep learning model 116B. In yet another instance, the circuitry 202 may apply the deep learning model 116N and may process multimedia content based on the application of the deep learning model 116N. Details related to the application of the selected deep learning model are described further, for example, in FIG. 3.

[0152] At 916, a verification result including a deep-fake indicator of the media content 110 may be generated. The circuitry 202 may, based on the application of the deep learning model 116A on the media content 110, may analyze the media content 110 to generate a verification result. In an instance, the verification result may include a deepfake indicator of the media content 110. Further, the generated verification result may be stored in the distributed ledger 112. Details related to the generation of the verification result are described further, for example, in FIG. 3.

[0153] At 918, first content associated with the media content 110 may be received from the trusted-content database 106B, based on the verification result. In some embodiments, the circuitry 202 may be configured to receive the first content through the network interface 206. Details related to the reception of first content are described further, for example, in FIG. 3.

[0154] At 920, the verification result may be stored on the distributed ledger 112, based on the content-based identifier and the first content. In some embodiments, the circuitry 202 may be configured to store the verification result on the blockchain network 308.Details related to the storage of the verification result are described further, for example, in FIG. 3 and FIG. 4.

[0155] At 922, a display device may be controlled. The circuitry 202 may be configured to control a display device (for instance, the user device 118) associated with at least one of a user or an owner associated with the media content 110, to render the deep-fake indicator of the media content 110. The circuitry 202 may render the deep-fake indicator of the media content 110 based on the stored verification result. Details related to the control of rendering are described further, for example, in FIG. 3. Control may pass to end.

[0156] Although the flowchart 900 is illustrated as discrete operations, such as, 904, 906, 908, 910, 912, 914, 916, 918, 920, and 922, the disclosure is not so limited. Accordingly, in certain embodiments, such discrete operations may be further divided into additional operations, combined into fewer operations, or eliminated, depending on the implementation without detracting from the essence of the disclosed embodiments.

[0157] Various embodiments of the disclosure may provide a non-transitory computer- readable medium and / or storage medium having stored thereon, computer-executable instructions by a machine and / or a computer to operate an electronic device (for example, the electronic device 102 of FIG. 1 ). Such instructions may cause the electronic device 102 to perform operations that may include receiving media content (for example, the media content 110 of FIG. 1 ) and first information including at least one of a content type or a content complexity associated with the media content 110. The operations may further include determining a content-based identifier associated with the media content 110. The operations may further include storing the content-based identifier on a distributed ledger (e.g., the distributed ledger 112). The operations may further include applying an executable code associated with the distributed ledger 112 on the first information, based on the stored content-based identifier. The operations may further include selecting, from a set of deep learning models (e.g., the set of deep learning models 116), a first deeplearning model based on the applied executable code. The operations may further include application of the selected first deep learning model on the media content. The operations may further include generation of a verification result including a deep-fake indicator of the media content, based on the application of the selected first deep learning model. The operations may further include receiving, from a trusted-content database (e.g., the trusted-content database 106B), first content associated with the media content, based on the verification result. The operations may further include storing the verification result on the distributed ledger 112, based on the content-based identifier and the first content. The operations may further include control of a display device (e.g., the user device 118) associated with at least one of a user or an owner associated with the media content, to render the deep-fake indicator of the media content, based on the stored verification result.

[0158] Exemplary aspects of the disclosure may provide an electronic device (such as, the electronic device 102 of FIG. 1 ) that includes circuitry (such as, the circuitry 202). The circuitry 202 may be configured to receive media content (for example, the media content 110 of FIG. 1 ) and first information including at least one of a content type or a content complexity associated with the media content 110. The circuitry 202 may be further configured to determine a content-based identifier associated with the media content 110. The circuitry 202 may be further configured to store the content-based identifier on a distributed ledger (e.g., the distributed ledger 112). The circuitry 202 may be further configured to apply an executable code associated with the distributed ledger 112 on the first information, based on the stored content-based identifier. The circuitry 202 may be further configured to select, from a set of deep learning models (e.g., the set of deep learning models 116), a first deep learning model based on the applied executable code. The circuitry 202 may be further configured to apply the selected first deep learning model on the media content 110. The circuitry 202 may be further configured to generate a verification result including a deep-fake indicator of the media content 110, based on theapplication of the selected first deep learning model. The circuitry 202 may be further configured to receive, from a trusted-content database (e.g., the trusted-content database 106B), first content associated with the media content, based on the verification result. The circuitry 202 may be further configured to store the verification result on the distributed ledger 112, based on the content-based identifier and the first content. The circuitry 202 may be further configured to control the user device 118 associated with at least one of a user or an owner associated with the media content, to render the deep-fake indicator of the media content, based on the stored verification result.

[0159] The circuitry 202 may be configured to receive the media content 110 of various content types corresponding to at least one of text content, audio content, image content, video content, augmented reality / virtual reality (AR / VR) content, or multi-media content.

[0160] The circuitry 202 may be configured to generate a hash value of the media content 110. The determination of the content-based identifier may be based on the generated hash value. The circuitry 202 may be further configured to store the media content on a decentralized storage and associate the media content 110with the contentbased identifier stored on the distributed ledger 112.

[0161] The circuitry 202 may be configured to receive second information including at least one of a source of the media content, a timestamp associated with the media content, or metadata associated with the media content, and store the second information on the distributed ledger 112 and associate the second information with the content-based identifier.

[0162] The circuitry 202 may be configured to select a second deep learning model from the set of deep learning models 116 based on the applied executable code. The circuitry 202 may further be configured to apply the selected second deep learning model on the media content 110. The generation of the verification result may further be based on the application of the selected second deep learning model. .

[0163] The circuitry 202 may be configured to determine a confidence score associated with the verification result. The circuitry 202 may further be configured to store the confidence score on the distributed ledger 112 and associate the confidence score with the verification result. The circuitry 202 may further be configured to control the user device 118 to render the confidence score associated with the verification result to at least one of a user or an owner associated with the media content 110.

[0164] The circuitry 202 may be configured to receive, from the trusted-content database 106B, second content associated with the media content. The circuitry 202 may further be configured to compare the first content and the second content. The circuitry 202 may further be configured to update the verification result based on the comparison.

[0165] The circuitry 202 may be configured to receive a user input based on the control of the rendered deep-fake indicator. The circuitry 202 may further be configured to update the verification result stored on the distributed ledger 112 based on the received user input.

[0166] The circuitry 202 may be configured to generate a notification including the verification result and may further transmit the notification to a second electronic device, for instance the user device 118, associated with the at least one of the user or the owner.

[0167] The circuitry 202 may be configured to receive an update to the set of deep learning models 116. The circuitry 202 may further be configured to re-select a deep learning model from the update to the set of deep learning models 116. The circuitry 202 may further be configured to apply the re-selected deep learning model on the media content 110. The circuitry 202 may further be configured to update the verification result based on the application of the re-selected deep learning model.

[0168] The circuitry 202 may be configured to receive a request to access the media content 110. The circuitry 202 may further be configured to verify the request based on the stored verification result. The circuitry 202 may further be configured to control access to the media content 110 based on the verification of the request.

[0169] The circuitry 202 may be configured to generate a distributed ledger transaction including the verification result. The circuitry 202 may further be configured to broadcast the distributed ledger transaction to the network of nodes 112A associated with the distributed ledger 112.

[0170] The present disclosure may be realized in hardware, or a combination of hardware and software. The present disclosure may be realized in a centralized fashion, in at least one computer system, or in a distributed fashion, where different elements may be spread across several interconnected computer systems. A computer system or other apparatus adapted for carrying out the methods described herein may be suited. A combination of hardware and software may be a general-purpose computer system with a computer program that, when loaded and executed, may control the computer system such that it carries out the methods described herein. The present disclosure may be realized in hardware that includes a portion of an integrated circuit that also performs other functions. It may be understood that, depending on the embodiment, some of the steps described above may be eliminated, while other additional steps may be added, and the sequence of steps may be changed.

[0171] The present disclosure may also be embedded in a computer program product, which includes all the features that enable the implementation of the methods described herein, and which when loaded in a computer system is able to carry out these methods. Computer program, in the present context, means any expression, in any language, code or notation, of a set of instructions intended to cause a system with an information processing capability to perform a particular function either directly, or after either or both of the following: a) conversion to another language, code or notation; b) reproduction in a different material form. While the present disclosure has been described with reference to certain embodiments, it will be understood by those skilled in the art that various changes may be made, and equivalents may be substituted without departing from the scope of thepresent disclosure. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the present disclosure without departing from its scope. Therefore, it is intended that the present disclosure is not limited to the particular embodiment disclosed, but that the present disclosure will include all embodiments that fall within the scope of the appended claims.

Claims

CLAIMSWhat is claimed is:1 . An electronic device, comprising: circuitry configured to: receive media content and first information including at least one of a content type or a content complexity associated with the media content; determine a content-based identifier associated with the media content; store the content-based identifier on a distributed ledger; apply an executable code associated with the distributed ledger on the first information, based on the stored content-based identifier; select, from a set of deep learning models, a first deep learning model based on the applied executable code; apply the selected first deep learning model on the media content; generate a verification result including a deep-fake indicator of the media content, based on the application of the selected first deep learning model; receive, from a trusted-content database, first content associated with the media content, based on the verification result; store the verification result on the distributed ledger, based on the contentbased identifier and the first content; and control a display device associated with at least one of a user or an owner associated with the media content, to render the deep-fake indicator of the media content, based on the stored verification result.

2. The electronic device according to claim 1 , wherein the set of content types corresponds to at least one of text content, audio content, image content, video content, augmented reality / virtual reality (AR / VR) content, or multi-media content.

3. The electronic device according to claim 1 , wherein the circuitry is further configured to generate a hash value of the media content, wherein the determination of the content-based identifier is based on the generated hash value.

4. The electronic device according to claim 1 , wherein the circuitry is further configured to store the media content on a decentralized storage and associate the media content with the content-based identifier stored on the distributed ledger.

5. The electronic device according to claim 1 , wherein the circuitry is further configured to: receive second information including at least one of a source of the media content, a timestamp associated with the media content, or metadata associated with the media content; and store the second information on the distributed ledger and associate the second information with the content-based identifier.

6. The electronic device according to claim 1 , wherein the circuitry is further configured to: select a second deep learning model from the set of deep learning models based on the applied executable code; apply the selected second deep learning model on the media content, whereinthe generation of the verification result is further based on the application of the selected second deep learning model.

7. The electronic device according to claim 1 , wherein the circuitry is further configured to: determine a confidence score associated with the verification result; store the confidence score on the distributed ledger, and associate the confidence score with the verification result; and control the display device to render the confidence score associated with the verification result.

8. The electronic device according to claim 1 , wherein the circuitry is further configured to: receive, from the trusted-content database, second content associated with the media content; compare the first content and the second content; and update the verification result based on the comparison.

9. The electronic device according to claim 1 , wherein the circuitry is further configured to: receive a user input based on the control of the rendered deep-fake indicator; and update the verification result stored on the distributed ledger based on the received user input.

10. The electronic device according to claim 1 , wherein the circuitry is further configured to: generate a notification including the verification result; and transmit the notification to a second electronic device associated with the at least one of the user or the owner.

11. The electronic device according to claim 1 , wherein the circuitry is further configured to: receive an update to the set of deep learning models; re-select a deep learning model from the update to the set of deep learning models; apply the re-selected deep learning model on the media content; and update the verification result based on the application of the re-selected deep learning model.

12. The electronic device according to claim 1 , wherein the circuitry is further configured to: receive a request to access the media content; verify the request based on the stored verification result; and control access to the media content based on the verification of the request.

13. The electronic device according to claim 1 , wherein the circuitry is further configured to: generate a distributed ledger transaction including the verification result; and broadcast the distributed ledger transaction to a network of nodes associated with the distributed ledger.

14. A method, comprising: in an electronic device: receiving media content and first information including at least one of a content type or a content complexity associated with the media content; determining a content-based identifier associated with the media content; storing the content-based identifier on a distributed ledger; applying an executable code associated with the distributed ledger on the first information, based on the stored content-based identifier; selecting, from a set of deep learning models, a first deep learning model based on the applied executable code; applying the selected first deep learning model on the media content; generating a verification result including a deep-fake indicator of the media content, based on the application of the selected first deep learning model; receiving, from a trusted-content database, first content associated with the media content, based on the verification result; storing the verification result on the distributed ledger, based on the contentbased identifier and the first content; and controlling a display device associated with at least one of a user or an owner associated with the media content, to render the deep-fake indicator of the media content, based on the stored verification result.

15. The method according to claim 14, further comprising: receiving second information including at least one of a source of the media content, a timestamp associated with the media content, or metadata associated with the media content; andstoring the second information on the distributed ledger and associating the second information with the content-based identifier.

16. The method according to claim 14, further comprising: determining a confidence score associated with the verification result; storing the confidence score on the distributed ledger, and associating the confidence score with the verification result; and controlling the display device to render the confidence score associated with the verification result.

17. The method according to claim 14, further comprising: receiving, from the trusted-content database, second content associated with the media content; comparing the first content and the second content; and updating the verification result based on the comparison.

18. The method according to claim 14, further comprising: receiving a user input based on the rendered deep-fake indicator; and updating the verification result stored on the distributed ledger based on the received user input.

19. The method according to claim 14, further comprising: receiving a request to access the media content; verifying the request based on the stored verification result; and controlling access to the media content based on the verification of the request.

0. A non-transitory computer-readable medium having stored thereon, computerexecutable instructions that when executed by an electronic device, causes the electronic device to execute operations, the operations comprising: receiving media content and first information including at least one of a content type or a content complexity associated with the media content; determining a content-based identifier associated with the media content; storing the content-based identifier on a distributed ledger; applying an executable code associated with the distributed ledger on the first information, based on the stored content-based identifier; selecting, from a set of deep learning models, a first deep learning model based on the applied executable code; applying the selected first deep learning model on the media content; generating a verification result including a deep-fake indicator of the media content, based on the application of the selected first deep learning model; receiving, from a trusted-content database, first content associated with the media content, based on the verification result; storing the verification result on the distributed ledger, based on the contentbased identifier and the first content; and controlling a display device associated with at least one of a user or an owner associated with the media content, to render the deep-fake indicator of the media content, based on the stored verification result.

Images