Tech Giants Urge US to Protect Open-Weight AI

Open-weight models let developers download parameters, deploy systems on their own infrastructure, and inspect or modify how they work, unlike closed models accessed mainly through a provider. Supporters say that flexibility lowers costs and technical barriers for startups, researchers and public agencies, while reducing dependence on a small group of AI vendors. The debate matters because Washington is weighing how to preserve US technological leadership without weakening safeguards around security and intellectual property.

On July 24, 2026, Nvidia Chief Executive Jensen Huang posted “Open Weights and American AI Leadership” on X. The initial 25 signatories included Microsoft, Meta, IBM, Dell Technologies, Hugging Face and Palantir, with OpenAI joining later. The letter urged US policymakers to avoid premature restrictions that could curb competition or push innovation overseas, expand access to computing and shared evaluation resources, and address genuine technology theft through targeted legal and commercial frameworks rather than broad limits on open weights or model distillation.