Anthropic refuses to sign open-weight AI letter, faces criticism from Silicon Valley peers

The US tech industry is rallying behind open-weight AI, but Anthropic faces widespread criticism for not signing a public letter supporting it. Last Friday, over 20 companies including Nvidia, Meta, and Microsoft signed the letter, later joined by OpenAI, Google, and SpaceX. As of Monday, Anthropic is the only top AI lab not backing the initiative.

Venture investor David Sacks and Benchmark partner Bill Gurley criticized Anthropic for refusing to sign, suggesting it's protecting its business interests. Anthropic CEO Dario Amodei previously argued that releasing model weights could lose control over cutting-edge AI, favoring controlled access.

Chinese AI company Moonshot AI released its open-weight model Kimi K3, which performs close to top US closed models, drawing US government attention. US Treasury Secretary Scott Bessent said sanctions could follow if Chinese models are trained using US models through distillation. Nvidia launched the 'Open Secure AI Alliance' to promote open AI in cybersecurity.