Researchers propose game theory method to fairly credit AI content creators

A new paper from researchers Keegan Harris, Siddharth Prasad, and Asher Trockman tackles the growing problem of fairly compensating creators when AI generates content—such as code, news articles, or short videos—using their intellectual property. Their proposed mechanism, In-Context Credit Assignment via the Core, draws on cooperative game theory. Specifically, it leverages the least core solution concept, which distributes value in the most stable way possible by ensuring no subset of creators is significantly under-compensated relative to the value they could generate independently. This prevents exploitation even when many parties contribute context.

To make this practical, the team developed novel algorithmic routines for constraint seeding and constraint separation. On a web retrieval credit assignment task, their methods approximated the least core using orders of magnitude fewer calls to large language models compared to alternative approaches. This efficiency leap makes the concept viable for real-world applications, like pay-per-use licensing or attribution systems for AI-generated outputs. The work sits at the intersection of AI, game theory, and machine learning, offering a principled path toward equitable creator compensation in an AI-driven economy.