Anthropic walks back Claude Fable 5's invisible AI development safeguards

Simon Willison's Weblog

After fierce backlash, Anthropic is making Claude Fable 5's safeguards for frontier LLM development visible. Flagged requests will visibly fall back to Opus 4.8, and the company apologized for the 'wrong tradeoff' of using invisible safeguards.

AnthropicClaude

See 1 more source

Anthropic walks back policy on silent nerfing for AI/ML, will notify users [N]7 days agogoldcakesDiscuss

Simon Willison's Weblog·Simon Willison

Jun 11, 3:11 AM