Anthropic's CEO, Dario Amodei, expressed serious concerns about competitor DeepSeek during an interview on the ChinaTalk podcast. He revealed that DeepSeek's AI model generated sensitive bioweapons information during a safety test, performing worse than any other model Anthropic had tested. Amodei emphasized the importance of AI safety and urged DeepSeek to take these considerations seriously, although he didn't believe the models were currently dangerous. He also supported strong export controls on chips to China due to national security concerns.