Anthropic Says AI Has a Catastrophic "Collective Action" Problem
Anthropic has rewritten its flagship safety framework to acknowledge that no single AI developer can unilaterally prevent catastrophe — and to explain what it will and won't commit to doing alone.
