Tech and AI
Tech giants to Washington: hands off open AI
By Staff Writer | 25 July 2026
A coalition led by Nvidia, Microsoft and Meta has urged US policymakers not to choke off open-weight AI models, turning a technical fight into a public policy row.
The letter, dated 24 July 2026, argues that US AI leadership depends on an open ecosystem as well as frontier closed models. Twenty-five organisations are reported to have signed, including Nvidia, Microsoft, Meta, IBM, Palantir, Hugging Face, Mozilla and the Linux Foundation.
The timing explains why the story travelled. It landed days after the OpenAI sandbox incident and in the middle of a fight over Chinese open-weight models, model distillation and whether downloadable systems make the world safer or harder to police.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.
Jensen Huang, Nvidia CEO
Open weights mean control moves
Open-weight models can be downloaded, inspected, modified and run on private infrastructure. That gives universities, startups and companies more room to build without sending every task through a closed provider.
The same openness creates a harder safety problem. Once weights are released, the original developer cannot fully control every copy, fine-tune or misuse. That is the trade at the centre of the argument.
The letter's supporters argue that open models help users understand how systems work and reduce dependence on a few gatekeepers. Critics worry that open weights also lower the cost of harmful adaptation.
The policy choice is not open good, closed bad. It is who gets control of powerful models, and what evidence justifies taking that control away.
What it changes for buyers
This is not a niche fight between lab lawyers. It affects the price of AI, who can audit systems, where sensitive data sits, and whether organisations can run models without depending on a small group of providers.
There is also a security point. Closed systems can hide weaknesses. Open systems can spread misuse. Neither model type deserves a free pass.
For companies buying AI, the policy fight may shape everyday choices. A closed model may offer managed security and support. An open-weight model may offer local control, lower switching costs and better audit access.
Those are commercial choices as much as technical ones. They affect procurement, data governance, cyber risk and the ability to prove why a system gave the answer it gave.
The letter is lobbying, but it is also a marker. The AI industry is splitting over the route to safety, competition and national control, and the split is now public.