Ai2’s Recommendations to OSTP to enable open-source innovation with the U.S. AI Action Plan

We at the Allen Institute for AI (Ai2) have submitted a recommendation to the Office of Science and Technology Policy (OSTP) in response to the White House’s Request for Information on an AI Action Plan. Our recommendations are singularly focused on enabling America to capture the numerous benefits of powerful AI and ubiquitous open-source AI systems. The foundation of modern machine learning successes are built on a history of open collaboration across institutions.

Our recommendations are:

1. Invest in a truly open ecosystem of innovation on American soil by funding multiple types of institutions, fostering cross-domain collaboration and sharing of essential AI development artifacts (e.g. models, data, code, and more), and building the strongest AI research and development capacity in the United States for the benefit of all Americans.

2. Avoid future burdensome requirements on open-source AI by steering regulatory bodies away from interventions that unintentionally hamper progress on developing AI. AI is a rapidly evolving technology and the best use cases for AI are only just now coming into focus. Over-regulating AI domestically means we may never see fully optimized AI use cases emerge in America; instead, those breakthroughs will happen overseas. without American input, values, or oversight.

2025 has been a landmark year for open-weight and truly open language models – we hope to see that continue. Open-source AI has the clear potential to make AI more beneficial in terms of quality and breadth of adoption. Realizing the future of beneficial AI isn’t a given, and today we are at an inflection point where the United States has the opportunity to choose its destiny. We believe that securing this future depends on broad and unwavering support for open-source efforts and open innovation in AI.