Digital Signature Matching for Proprietary Code Leakage Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Publicly-accessible large language models (LLMs) pose privacy and security risks by potentially disclosing proprietary source code and internal contributions, making it challenging to identify and protect intellectual property.
Innovation Solution
Utilizing digital signatures generated through piecewise hashing, Merkle trees, Simhash, Minhash, and content defined chunking to identify proprietary content, and implementing a computing platform with a signature generator, matching engine, and action tool to perform automated actions based on configurable policies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If digital signatures are used to identify proprietary content in user submissions to LLMs, then intellectual property protection is improved, but computational resources and storage needs increase
Solution Approach 1:
The patent segments proprietary content into distinct components (text, code, images, audio, video) and generates separate digital signatures for each type. This segmentation allows the system to process and protect only the relevant content types present in each submission, rather than uniformly processing all content, thereby reducing unnecessary computational overhead while maintaining comprehensive IP protection.
Solution Approach 2:
The system implements partial action by selectively applying digital signature generation and content analysis based on the specific type and sensitivity of content detected. Rather than analyzing every byte of every submission equally, the system applies appropriate protection mechanisms only where needed, balancing thorough IP protection with efficient resource utilization.
2Measurement precision
If comprehensive content analysis is performed to identify all proprietary information, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent applies local quality by using different analysis methods and signature generation approaches tailored to specific content types. Text content receives linguistic analysis, code receives syntax-aware processing, and multimedia content receives appropriate format-specific handling. This localized approach ensures high identification accuracy for each content type while avoiding the application of overly complex methods uniformly across all content.
Solution Approach 2:
The system introduces intermediary components (content type detectors, signature generators, and matching engines) that facilitate the complex analysis process. These intermediaries break down the overall complex task into manageable stages, where each intermediary handles a specific aspect of content analysis, thereby making the overall system more tractable and maintainable while achieving comprehensive detection accuracy.
3Reliability
If automated actions are implemented to block transmissions of proprietary content, then intellectual property protection is improved, but ease of operation deteriorates
Solution Approach 1:
The system implements feedback mechanisms that notify users when their submissions contain proprietary content and explain the reasons for blocking. This feedback loop allows users to understand what triggered the protection mechanism, review their content, and resubmit corrected versions if appropriate. The feedback approach maintains strong IP protection while preserving user autonomy and ease of operation through transparent communication.
4Reliability
If digital signatures are generated for all content types, then intellectual property protection is improved, but loss of time increases
Solution Approach 1:
The system performs preliminary actions by pre-generating and storing digital signatures for proprietary content in a database before actual submissions occur. When user content is submitted, the system quickly compares against the pre-computed signatures rather than generating new signatures in real-time. This preliminary preparation significantly reduces processing time during actual content submission while maintaining comprehensive protection capabilities.
Data Source
AI summary
Methods of protecting intellectual property using digital signatures include generating reference digital signatures of first content, generating a digital signature of second content, comparing the digital signature of the second content to the reference digital signatures to identify matching reference digital signatures, and selectively performing an action based on the matching reference digital signatures and a policy. The first content may include proprietary information of an organization and/or posts of a machine-learning (ML) model. The second content may include source code intercepted from a transmission directed to an external site, such as a ML model, and/or source code saved to a source code repository. Actions may include, without limitation, initiating an audit of the second content, sending a notification to a user interface indicating that the second content likely contains a portion of the first content, releasing the transmission, and/or terminating the transmission.