Local Resource Authentication Using AI-Generated Distractor Prompts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Single-factor authentication provides inadequate protection against misuse and malicious intrusion, necessitating improved cybersecurity measures for computer resource access.
Innovation Solution
A system leveraging local and cloud user resources, combined with generative artificial intelligence, creates tailored prompts using machine learning to authenticate users by presenting artificial distractors, ensuring correct resource selection grants access while incorrect selections deny access and trigger security actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If single-factor authentication is used, then ease of operation is improved, but security protection is worsened
Solution Approach 1:
The system creates synthetic copies of user resources (images, audio, video) that are indistinguishable from real user resources. These synthetic resources are presented as distractors during authentication, forcing the user to identify the actual real resource among fake copies. This resolves the contradiction by maintaining ease of operation through resource-based authentication while enhancing security through the difficulty of distinguishing real from synthetic resources.
Solution Approach 2:
The system dynamically changes the parameters of authentication by selecting different resource types (images, audio, video) and generating synthetic versions with varying characteristics. The synthetic resources are created using generative AI models that can modify parameters such as visual features, audio characteristics, or video properties to create convincing distractors. This parameter transformation enables multi-factor authentication while maintaining user-friendly interaction.
2Reliability
If multi-factor authentication is implemented, then security protection is improved, but device complexity is worsened
Solution Approach 1:
The system uses a universal resource-based authentication framework that can handle multiple factor types through a single cohesive mechanism. Instead of implementing separate authentication systems for different factors, the platform uses generative AI to create synthetic versions of any user resource (images, audio, video, documents), providing multi-factor authentication capabilities through a unified resource-based approach. This universality reduces overall system complexity while maintaining enhanced security.
Solution Approach 2:
The system introduces synthetic resources as an intermediary element between the user and the authentication system. Rather than directly comparing multiple different factor types, the synthetic resources serve as a mediator that transforms various authentication factors into a common comparison framework. This intermediary approach simplifies the authentication process while enabling multi-factor verification through resource similarity assessment.
3Measurement precision
If generative AI and machine learning models are used, then authentication accuracy is improved, but use of energy is worsened
Solution Approach 1:
The system applies partial action by selectively generating synthetic resources only when needed for authentication, rather than continuously running complex generative AI models. The machine learning models are activated partially - only during the authentication challenge phase - and the system uses excessive computational resources temporarily during model generation, then shuts down or reduces activity during normal operation. This partial activation approach maintains high authentication accuracy when required while significantly reducing average energy consumption.
Data Source
AI summary
Embodiments relate to performing user validation using local resources in order to permit or deny access to computer resources. An aspect includes in response to receiving a request for access, selecting a selected resource type from a plurality of resource types and selecting a selected user resource from a user resource pool, the selected user resource having the selected resource type. An aspect includes executing a machine learning model to output a prompt in response to inputting the selected user resource, executing a generative artificial intelligence (AI) engine to output generated resources in response to inputting the prompt, and performing an authentication by presenting the generated resources and the selected user resource.