The emergence of generative foundational large language models has put the focus on the responsible use of these models by others. New forms of open source licenses for AI models have emerged that aim to encourage responsible use of AI [1], including conformance to the obligations in high risk application that will be regulated under the forthcoming EU AI Act. These are starting to be used at open AI model repositories such as [2]. However, currently such licenses are not accompanied by any oversight mechanism beyond taking irresponsible users to court under copyright law.
This project will explore how such repositories could operate as a shared dataspace where access to models could be governed by machine readable agreements using the Open Digital Rights Language [3] which could operate as smart contracts requiring reporting obligations to demonstrate responsible use by model downloaders.