LLM Authorization Proxy for Context-Aware Data Access
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems fail to effectively constrain and tailor the output of Large Language Models (LLMs) within organizational contexts, leading to unauthorized access to sensitive information, as they are not tailored to the specific roles and broader organizational context of querying users.
Innovation Solution
Implement a pre-defined Selective LLM Authorization Policy that dynamically tailors LLM outputs based on user-related and data-related organizational context, using grounding techniques and prompt modifiers to filter or adapt responses according to user permissions and organizational data access rules.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If LLM is used to query organizational data lake, then information accessibility is improved, but data security and compliance deteriorate due to unauthorized access to sensitive information
Solution Approach 1:
The patent introduces an authorization proxy unit as an intermediary component between the user query and the LLM system. This proxy unit intercepts queries, determines user authorization levels, and modifies prompts accordingly before they reach the LLM, thereby preventing unauthorized access to sensitive information while maintaining information accessibility for authorized users
Solution Approach 2:
The system applies different authorization policies and prompt modifications to different users based on their specific roles, departments, and clearance levels. Each user receives a customized prompt that reflects their unique authorization context, allowing tailored information access control rather than a one-size-fits-all approach
2Device complexity
If uniform LLM output is provided to all users, then system simplicity is maintained, but organizational context and user-specific constraints are lost
Solution Approach 1:
The system dynamically modifies LLM prompts based on real-time authorization context, user roles, and query content. The authorization proxy unit adjusts prompt parameters, constraints, and data access permissions dynamically for each user interaction, enabling the system to adapt to organizational context without requiring complex manual configuration
Solution Approach 2:
The system changes prompt parameters such as temperature, max tokens, and data access permissions based on user authorization levels and organizational policies. By modifying these parameters dynamically, the system maintains operational simplicity while achieving context-aware behavior through parameter adjustment rather than structural complexity
Data Source
AI summary
A computerized system receives an original prompt that a querying user sends to a Large Language Model (LLM) that is operably connected to organizational data sources of an organization. Instead of executing the original prompt by the LLM, the system obtains user-related organizational context that pertains to characteristics of the querying user, obtains data-related organizational context that pertains to data from which the LLM is expected to obtain information for responding to the original query, and obtains pre-defined organizational policy rules, that indicate which type of users are authorized to access which type of organizational data. Based on the obtained data, the system modifies the original prompt into an adapted prompt. The system sends the adapted prompt, and not the original prompt, to the LLM for processing. The system obtains LLM-generated output from the LLM in response to the adapted prompt, and provides that LLM-generated output to the querying user.