Blockchain Voice and Face Bank for AI Content Fraud Prevention
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The authenticity of multimedia content, such as videos and audio, is difficult to verify due to the ease of manipulation using AI technologies like deepfakes, which can alter faces, voices, and actions, making it challenging to confirm the validity of events and spoken words.
Innovation Solution
A system utilizing a distributed ledger and blockchain to register scheduled events, generate and embed modulation schemes in multimedia content, making it harder to alter and allowing for authentication by matching these schemes, thereby ensuring the content's integrity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional voice matching and face matching techniques are used to authenticate multimedia content, then the authentication process can identify whether a video is from the actual person, but it is unable to verify the content against a real event and confirm its authenticity
Solution Approach 1:
The system performs preliminary registration of scheduled events before they occur, creating a reference database of authentic content. This advance preparation enables subsequent verification to compare new content against pre-established authentic records, ensuring both identification accuracy and event authenticity.
Solution Approach 2:
The patent introduces an intermediary verification system that acts as a mediator between the multimedia content and the authentication process. This intermediary layer compares content against multiple reference factors (voice, face, event context) to provide comprehensive authenticity verification beyond simple matching.
2Productivity
If AI technologies are used to generate multimedia content, then content creation becomes more efficient and accessible, but it enables easy manipulation and creation of deepfakes that look convincingly real
Solution Approach 1:
The system applies preliminary anti-action by embedding authentication markers and metadata in legitimate multimedia content before distribution. This preemptive measure creates a detectable signature that can later identify authentic content versus AI-generated deepfakes, countering the harmful effects of manipulation.
Solution Approach 2:
The patent implements a feedback mechanism where the verification system provides authentication results that can be used to improve detection algorithms. By analyzing both authentic and manipulated content, the system learns to better distinguish between real and fake media, continuously improving its ability to counter deepfakes.
3Speed
If multimedia content is distributed over the Internet through social media networks, then content reach and distribution speed increase, but the authenticity of the content cannot be guaranteed and can be easily altered and re-distributed
Solution Approach 1:
The system performs preliminary authentication and tagging of content before it enters the distribution network. By embedding verification data in advance, the content carries its authenticity proof through the entire distribution chain, enabling verification at any point without slowing down the rapid sharing characteristic of social media.
Solution Approach 2:
The patent uses cryptographic copying techniques where an immutable hash or digital signature of the original content is created and distributed with the content itself. This creates a verifiable copy that proves the content's origin and integrity, allowing fast distribution while maintaining authenticity verification capability.
Data Source
AI summary
A method for processing multimedia content for authentication and authenticating the multimedia content includes registering a scheduled event, creating a mapping between the scheduled event and a registrant of the scheduled event and in response to creating the mapping between the scheduled event and the registrant, issuing a request to add a block in a blockchain for the mapping. The method also includes in response to the request, receiving, a block identifier, generating a start code for one or more modulation schemes based on the block identifier, receiving multimedia content generated based on the scheduled event and modifying the multimedia content based on the one or more modulation schemes for distribution of the multimedia content.