Open Letters on AI Development Proliferate

A series of open letters concerning open-weight models and the pace of frontier AI development have emerged in the AI industry.

• The Microsoft-led ‘Open Weights and American AI Leadership’ letter was signed by 235 companies, including NVIDIA, Amazon, Y Combinator, Linux Foundation, and OpenAI.
• The letter argued that relying solely on closed models could increase single points of failure and market concentration, and voiced support for open-weight models that allow researchers and developers to verify model behavior and improve safeguards.
• In particular, it argued that distillation—using another model's outputs for training—should be recognized as a legitimate model development and evaluation technique.

Anthropic did not join the signatories and instead published a separate position. While not calling for an outright ban on open-weight models themselves, it stated that industrial-scale distillation operations should be cracked down on, citing the risk of authoritarian states developing high-performance AI and the risk of cyber and biological attacks.

Another open letter, Pacing the Frontier, was signed by 1,324 employees of frontier AI companies, including figures from OpenAI, Anthropic, and Safe Superintelligence. The letter urged that, as automated AI research accelerates technological progress and competitive pressure, the U.S. government should establish technical and governance tools to deliberately regulate the pace of AI development on an international scale.