If powerful AI is coming regardless, the safest outcome is for it to be built by people who are genuinely worried about the risks.

safetymissionDario Amodei, interview

We left OpenAI because we believed safety shouldn't be a department — it should be the foundation everything is built on.

safetyopenaiDario Amodei, paraphrased from interviews

Constitutional AI is the idea that you can train a model to follow principles, not just examples. You teach it values, not just tricks.

constitutional-aisafetyDario Amodei, research paper summary

The most important thing about Claude is not what it can do — it's what it chooses not to do.

safetyproductDaniela Amodei, interview

We're in a race to build the most powerful technology in human history. The question isn't whether to race — it's whether to race responsibly.

AI-raceresponsibilityDario Amodei, essay

I'm optimistic about AI's potential to solve enormous problems. I'm also terrified of getting it wrong. Both feelings are appropriate.

optimismriskDario Amodei, podcast
Anthropic Quotes | Netfigo