Joe Carlsmith(@jkcarlsmith):In case one more data point is useful: I too am an AI company employee who thinks the risk of extinction-level catastrophe from rogue AI is >10%. I wrote about this back in 2022, and again in a 2025 post about why I was joining Anthropic (links in thread). I still believe it.

Joe Carlsmith

@jkcarlsmith

Philosophy, futurism, AI. Working on Claude's values @AnthropicAI. Formerly @coeff_giving. Opinions my own.

加入 April 2013

694 正在關注 13.2K 粉絲

Joe Carlsmith@jkcarlsmith

2026.09.10 23:18

In case one more data point is useful: I too am an AI company employee who thinks the risk of extinction-level catastrophe from rogue AI is >10%. I wrote about this back in 2022, and again in a 2025 post about why I was joining Anthropic (links in thread). I still believe it.