Two dozen companies and organizations have signed an open letter to US policymakers advocating for the continued availability of open-weight AI models.

A coalition of two dozen companies and organizations, including Meta, Microsoft, Nvidia, IBM, and Dell Technologies, has signed an open letter urging US policymakers to protect open-weight AI models. The signatories argue that these models, which allow users to download and modify trained parameters, are essential for fostering competition and innovation across the technology sector.

Arguments for Open-Weight Models

The letter contends that open-weight models allow AI capabilities to reach a wider range of industries, including hospitals, farms, and classrooms. By lowering the cost of entry, these models enable startups and public institutions to avoid reliance on expensive, closed-source frontier models. The signatories also highlight that open weights help enterprise customers avoid vendor lock-in, granting them control over their own data and model adaptations.

Security and Regulatory Stance

Addressing security concerns, the signatories argue that open models are not inherently less secure than closed systems. They suggest that open-weight models allow for broader red-teaming and vulnerability identification by outside researchers, comparing this to the open-source software movement. The letter also defends the practice of model distillation, urging policymakers to address potential misappropriation through targeted legal mechanisms rather than imposing broad restrictions on the technique.

Policy Implications

While the letter does not propose specific legislation, it serves as a positioning document ahead of anticipated AI policy discussions in Washington. It calls on lawmakers to fund shared training datasets, expand compute access for researchers, and avoid premature restrictions on open models. The document reflects the interests of infrastructure and chip providers who benefit from a diverse ecosystem of deployable AI models.