So the current line is: Anthropic and openai both believe that their product has a chance of wiping humans off the face of the earth. They both know that if they slowed down, they could probably understand the chances better, or maybe even reduce them to zero. They refuse to slow down, even though that risks the literal end of the world, because the other one might end up with the apocalypse machine instead of them. It’s just stupid enough to be true, that.