AI-Generated Media Authentication With Metadata-Based Login Challenges
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-factor authentication methods rely on secondary authentication tasks that are easy for users to remember but difficult for malicious actors to guess, posing a challenge in security and efficiency.
Innovation Solution
A computerized method using AI-generated media samples for authentication, where users select preferred samples during setup, and later authenticate by identifying these samples among newly generated ones, with the system storing metadata for sample generation, enhancing security and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If traditional secondary authentication questions are used, then users can easily remember the answers, but malicious actors can easily guess the answers
Solution Approach 1:
The patent uses AI-generated copies or variations of the user's preferred media samples as authentication challenges. Instead of using the exact original sample, the system generates multiple similar versions and presents them to the user, who must identify the correct one. This approach maintains ease of recognition for the legitimate user while preventing attackers from easily guessing the answer, since they cannot determine which variation is the correct copy.
2Reliability
If AI-generated media samples are used for authentication, then security is enhanced by making pattern recognition difficult, but the complexity of the authentication system increases
Solution Approach 1:
The system integrates multiple functions into a unified authentication framework: user preference collection, AI sample generation, challenge presentation, and verification all work together in a cohesive flow. The AI model serves multiple purposes - generating both the authentication challenges and the distractor options, while the preference data serves both personalization and security verification functions.
3Measurement precision
If full media samples are stored for authentication verification, then accurate matching can be performed, but data storage requirements increase significantly
Solution Approach 1:
The system extracts only the essential identifying features or metadata from the full media samples during the preference collection phase. These extracted characteristics are then used for verification against the AI-generated challenges, eliminating the need to store and process complete high-resolution media files while maintaining sufficient accuracy for authentication purposes.
Data Source
AI summary
A computerized method performs an authentication process using AI-generated media samples. A login request is received from a user, and, in response to the login request, a preferred sample metadata value associated with the user is obtained. The preferred sample metadata value is associated with a sample category. Using the preferred sample metadata value, a generative AI model is used to generate a preferred sample of the sample category. Further, the model is used to generate other samples of the sample category. The other samples do not include samples generated from the preferred sample metadata value. The generated preferred sample and the other samples are provided to the user via an interface and the user is prompted to select a preferred sample from the provided samples. When the user response to the prompt matches the provided preferred sample, the login request is granted to the user.