Meta, NVIDIA, Microsoft and Palantir ask regulators not to restrict open-weight formats
The coalition letter landed alongside Zuckerberg's 14-page manifesto, which asks Washington to remove policy friction around distillation and training data so American open models can compete. Meta also gave its independent board the power to approve safety criteria for model releases.
Four companies with very different exposure signed the same document. NVIDIA sells the hardware open models run on. Microsoft hosts them. Palantir deploys them into government. Meta makes them. The only interest they share is that open weights remain legal to publish and to fine-tune — which tells you what they think the actual risk is.
Zuckerberg's letter, titled "The Future is for Everyone: The Path to a Positive AI Future," rests on three claims: individual empowerment as the source of prosperity, invention as the primary purpose of superintelligence, and balance of power as the foundation of safety. The third is the load-bearing one. It reframes open weights from a distribution choice into an anti-concentration argument — the biggest risk is one entity with too much control, and the remedy is that many entities hold the weights.
The specific asks are narrower than the philosophy. Zuckerberg wants friction removed around distillation and data usage, both of which are live litigation questions rather than settled law. Asking regulators to bless distillation is asking them to bless training on competitors' outputs, and no American lab has an interest in that rule being clarified in Meta's favour except Meta.
The governance piece is the concession. An independent board that approves safety criteria and reviews whether individual releases meet them is a real constraint if it ever says no, and a press release if it does not. Meta has not published the criteria.
Axios — Zuckerberg: AI's biggest risk is one entity with too much control → · AI News — Zuckerberg details Meta's personal AI superintelligence strategy →