Amodei says Anthropic held Mythos back four months for safety
Anthropic CEO Dario Amodei defended the company's "race to the top" strategy in a post on X, listing nine decisions that cost it commercially but made AI safer. Mythos-class models were ready in February but only became generally available in June, after stronger cyber safeguards were in place.
- Mythos models were held from February to June, limited to cyber-defence partners
- Amodei admitted sacrificing a significant chunk of a several-month model lead
- Anthropic restricted sales to companies controlled by China
- A new public benefit fund will hold 5% of Anthropic's shares
Read next
Policy