Anthropic Updates Claude Biology Safeguards, Yet It Still Refuses Basic Questions
iamaliveix · x · 2026-08-08
Anthropic officially announced an update to Claude Fable 5's biology safeguards to reduce false positives, reportedly cutting biology-related fallbacks by 85%. However, user tests reveal the model still refuses to answer basic questions like "how babies are made."
The company stated that for dual-use requests (like virology, toxicology, and molecular design), Fable will continue to fall back to Opus 5, meaning it is not yet fully suitable for professional biological research and drug development.
More from Fun
- Dev Gives AI Agent $1000 and Free Will, It Names Itself Atlas — kleffew94 · 2026-08-08
- Google Interview: How to Improve a Vision Model? Candidate Says 'Sharingan' and Gets Rejected — prajdabre · 2026-08-08
- AI Safety Policy Program Launches with Hidden Prompt Injection Easter Egg — austinc3301 · 2026-08-08
- New AI Meme: The Guy Who Didn't Sign the NDA vs. The One Who Rejected the AGI Lab — JacquesThibs · 2026-08-08
- AI Safety Researcher Eric Mitchell Jokes: "We Shipped AGI!" — ericmitchellai · 2026-08-08
- AI Coding Tools Ranked by 'Vibes': Grok and Warp Top the List, Copilot NGMI — charlieholtz · 2026-08-08