MOT: A tool to fight openwashing in AI (lwn.net)

At the Open Source Summit North America 2026, Arnaud Le Hors introduced the Model Openness Tool (MOT), a vital resource aimed at tackling the issue of "openwashing" in AI, particularly concerning large language models (LLMs). Many models advertised as open source possess restrictive licenses that do not comply with the Open Source Initiative's standards, leading to misconceptions among users about their rights to use the models freely. MOT categorizes models into a tiered system based on their openness and completeness, helping users better assess the genuine accessibility of these resources. The significance of MOT lies in its potential to clarify what constitutes an "open" model and mitigate legal risks associated with misuse of LLMs. By establishing a classification scheme that ranks models from Class I ("Open Science Model") to Class III, MOT encourages model providers to enhance the openness of their offerings. Additionally, the tool serves as a registry for submitted models, enabling users to contribute to model evaluations while fostering a community-based approach to improving transparency in AI licensing. This initiative is part of broader efforts within the Generative AI Commons working group to establish responsible frameworks for the development and deployment of AI technologies.