Twenty-five technology companies, industry bodies and venture-capital firms have urged US policymakers to support open-weight AI models, arguing that regulation should not concentrate control of the market among a few large proprietary-model providers. The intervention matters because Washington is increasing scrutiny of AI following security concerns and recent restrictions, while lawmakers consider rules for a technology they are still trying to understand.

Signatories include Dell, IBM, Meta, Microsoft, Mistral, Mozilla, Nvidia, Palantir and Perplexity; Anthropic, Google and OpenAI did not sign, although OpenAI’s Sam Altman said he supported both open and proprietary models. The letter defines open-weight models as downloadable, modifiable systems that can run on users’ own infrastructure, but whose source code, training data and other artefacts may not be available for full independent reproduction; it argues they promote competition, confidence, safety and wider economic adoption.