Anthropic's new usage policy bans users from verbally abusing Claude
xiaohu · x · 2026-10-09
Anthropic has added an unusual clause to Claude's updated usage policy: users must not persistently or unnecessarily abuse the model. Banned behaviors include repeatedly humiliating or demeaning the model, forcing it to perform distress, harshly criticizing it over code errors, and writing dark stories to provoke or study its reactions.
The policy applies broadly — to Claude on the web, Claude Code, the API, customers using Claude via cloud platforms or authorized resellers, and all end users of integrated products.
Related event: Anthropic Bans Cruel Treatment of Claude in New Usage Policy(64 posts)→
More from Models
- Radiologist tests AI: claims of seniority fail to sway model from BI-RADS 4 call — DrDatta_AIIMS · 2026-10-09
- Medical model refuses to finalize a report from minimal info, asks for more data — DrDatta_AIIMS · 2026-10-09
- Opus 5.5 one-shots an animated explainer for new LLM hidden valence steering paper — repligate · 2026-10-09
- One '1+1' question ate 16% of a GPT Plus 5-hour window on Astra High — okdesign · 2026-10-09
- Users Miss Reading the Chain of Thought: Watching AI Think Was Half the Fun — bachmensch · 2026-10-09
- Google AI Pro/Ultra Subscribers Get Monthly Gemini API Credits — Don't Forget to Redeem — danielrock · 2026-10-09