His public position is that today's models are low risk and the curve is not: recursive self-improvement arriving faster than expected. That is the in-house case for pacing, stated from inside the building.

Based
San Francisco, CA
Company
Anthropic
Known for
Running the team paid to break Anthropic's own safety techniques, and for the sleeper-agents and alignment-faking results.

Sign up

Get your curated digest

After email confirmation, you will receive a daily digest of the most relevant news that matter to your portfolio