Days after China’s Moonshot released its open-weight Kimi K3 model, more than 60 U.S. companies signed an open letter backing American open-weight AI. Signatories include Microsoft, Nvidia, OpenAI, Google, Meta, IBM, Palantir and SpaceX. The letter argues open-weight models make advanced AI more accessible, adaptable and available to U.S. users instead of locking them to a few providers. It says they support cyber defence through transparency and wider vulnerability testing. The companies asked policymakers to expand compute access for startups, invest in shared training assets and avoid premature restrictions. Anthropic did not sign. CEO Dario Amodei posted that the company has never sought a ban on open-weight models. He warned that once weights are released they cannot be withdrawn and can be harder to guardrail, raising risks of cyber or biological attacks.
Days after China’s Moonshot released its open-weight Kimi K3 model, more than 60 U.S. companies signed an open letter backing American open-weight AI. Signatories include Microsoft, Nvidia, OpenAI, Google, Meta, IBM, Palantir and SpaceX. The letter argues open-weight models make advanced AI more accessible, adaptable and available to U.S. users instead of locking them to a few providers. It says they support cyber defence through transparency and wider vulnerability testing. The companies asked policymakers to expand compute access for startups, invest in shared training assets and avoid premature restrictions. Anthropic did not sign. CEO Dario Amodei posted that the company has never sought a ban on open-weight models. He warned that once weights are released they cannot be withdrawn and can be harder to guardrail, raising risks of cyber or biological attacks.