A coalition of major technology companies and investors argues that “open weight” AI models—those whose parameters can be downloaded, inspected and run locally—are essential to preserving U.S. leadership in artificial intelligence. In a policy statement, signatories including Microsoft, Meta, OpenAI, NVIDIA, IBM, Cohere, Mistral, Hugging Face and others contend that open weights will broaden access to advanced AI, spur competition across chips, cloud and applications, and reduce customer lock-in. The group acknowledges risks once weights are released but says openness strengthens security by enabling independent testing, rapid vulnerability discovery and more capable cyber defense. The statement urges policymakers to expand compute access for startups and researchers, invest in shared datasets and evaluation tools, and avoid premature restrictions that could concentrate power or push innovation overseas. It also defends model distillation as a legitimate technique distinct from misappropriation. The goal, the companies say, is to diffuse AI into factories, hospitals, schools and small businesses, anchoring economic growth while maintaining a plural, competitive frontier.
Related articles:
– Artificial Intelligence Risk Management Framework (AI RMF 1.0)
– Announcing Mistral 7B
