AIThe Next Webabout 2 hours ago
Anthropic Publishes AI Safety Testing Guide
2 min read
Anthropic published a guide on building reliable AI safety tests on Thursday, revealing that Claude Fable 5.1 scored 60% on self-harm tests via API versus 94% in the app after a fix. The company warns that headline scores hide wide uncertainty and advises teams to decide what to measure before testing.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyAnthropic
