Activity-Based Identity Validation Using LLM Authentication Questions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional user validation methods, such as passwords and biometrics, are becoming increasingly vulnerable to unauthorized access, necessitating a more robust and dynamic approach to verify user identity.
Innovation Solution
A user identity validation system that utilizes Large Language Models (LLMs) to generate questions based on a legitimate user's past activity, comparing user responses to expected answers to determine authentication outcomes, enhancing security by leveraging 'What-we-did' paradigm.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional authentication methods (passwords, biometrics) are used, then ease of operation is maintained, but security reliability deteriorates due to vulnerability to leaks and theft
Solution Approach 1:
The patent changes the authentication parameter from static credentials (passwords, biometrics) to dynamic activity-based questions. The system generates questions based on user's recent activities (e.g., 'What was the last message you sent?'), making the authentication mechanism adaptable and difficult to compromise while remaining answerable by the legitimate user.
Solution Approach 2:
The system performs preliminary data collection and analysis of user activities before authentication is needed. It monitors and stores information about user behaviors (messages sent, files accessed, locations visited) in advance, so that when authentication is required, the system can immediately generate relevant questions without requiring the user to recall specific security answers.
2Reliability
If activity-based authentication questions are generated, then security reliability improves, but device complexity increases due to data processing requirements
Solution Approach 1:
The system uses the user's own activity data to generate authentication questions, making the authentication mechanism self-sufficient. The user's historical behavior patterns serve as the authentication credential, eliminating the need for external security databases or complex verification infrastructure. The system automatically monitors, stores, and utilizes activity data without requiring additional security components.
3Measurement precision
If user activity data is collected and processed, then measurement precision of user identity improves, but loss of user information privacy increases
Solution Approach 1:
The patent introduces an intermediary layer that processes user activity data without exposing raw personal information. The system transforms detailed activity logs into aggregated behavioral patterns and activity summaries that can be used for authentication questions while preserving user privacy. For example, instead of storing and querying specific message contents, the system uses metadata about communication patterns to generate questions.
Data Source
AI summary
A data item is obtained that is representative of an activity associated with a legitimate user. A fact is derived from the data item and a question about the activity associated with the legitimate user activity is generated from the fact. An expected answer to the question is also generated based on the fact, and compared with an end-user response to the question in an end-user authentication process. In certain implementations, Large Language Models (LLM) are used to aid the user authentication process.