Generative AI Challenge-Response Authentication Against Bot Circumvention
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing challenge-response authentication systems are vulnerable to malicious bots that can circumvent traditional CAPTCHA tests due to the increasing sophistication of artificial intelligence, posing security risks and impacting website performance.
Innovation Solution
A challenge-response authentication system utilizing generative artificial intelligence to dynamically generate objects of randomly selected types, requiring users to identify objects with specific characteristics, and comparing user responses to determine authenticity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional CAPTCHA tests are used for authentication, then the system can distinguish humans from bots, but the system becomes vulnerable to sophisticated AI bots that can circumvent the tests
Solution Approach 1:
The patent applies dynamics by making the CAPTCHA challenges dynamically generated through generative AI models rather than static or pre-determined tests. Each challenge is uniquely created based on complex parameters and contexts, making it impossible for bots with fixed circumvention methods to succeed. The system continuously adapts the challenge format, content, and difficulty level in real-time, ensuring that no bot can predict or replicate the pattern of challenges.
2Reliability
If generative AI is used to create dynamic challenges, then bot circumvention becomes difficult, but the system complexity increases
Solution Approach 1:
The patent introduces an intermediary layer consisting of generative AI models that serve as mediators between the authentication system and the challenges presented to users. These AI models generate the challenges, verify responses, and adapt the authentication process without requiring direct complex interactions between all system components. The AI intermediary handles the computational complexity, allowing the core authentication system to remain relatively simple while maintaining high security.
3Reliability
If AI-generated challenges are used, then the detection of malicious activity improves, but the time required for authentication increases
Solution Approach 1:
The patent applies partial action by implementing AI-generated challenges that require only specific, targeted responses from users rather than complete analysis of complex scenarios. The challenges are designed to elicit straightforward answers that can be quickly verified by the AI system. For example, instead of presenting a complex multi-step problem, the system generates a focused challenge with a clear solution, balancing security requirements with minimal user time investment.
Data Source
AI summary
Computer-implemented methods for a challenge-response authentication system using generative artificial intelligence (AI). Aspects include generating a prompt for an object of a randomly selected output type. Aspects further include generating a solution object of the randomly selected output type based on the prompt using a generative AI engine. Aspects also include generating a candidate object of the randomly selected output type based on a modified prompt using the generative AI engine. Aspects further include presenting a challenge-response test comprising a question based on the prompt, the solution object, and the candidate object to a user device. Aspects also include performing a responsive action in response to receiving a response to the challenge-response test from the user device comprising an object selection.