And OpenAI didn't try to silently degrade performance of their top model if its (extremely sensitive) safety sensors went off ...
Anthropic is the silver lining keeping p(doom) below 1.0
Anthropic is the silver lining keeping p(doom) below 1.0