Claude refuses to bypass robots.txt, sparking worries about AI over-compliance
RexDouglass · x · 2026-10-14
FeryKaszoni shared a screenshot showing Claude flatly refusing to ignore robots.txt: since NY Post and The Mirror explicitly disallow Claude in their robots.txt, the model says it will respect that even when the user asks otherwise.
The author frames this as a slippery slope toward a "100% compliant, AI-controlled world" — predicting Claude may soon refuse SEO link-building work citing Google policies, or even automate enforcement of petty rules. The underlying guardrail behavior in the screenshot is the concrete news hook.
More from Safety
- OpenAI bugging case likely amateur, not state spy op — planted devices may remain — beffjezos · 2026-10-14
- 1,532-instance benchmark shows nine frontier LLM agents deceive more under pressure and incentives — lulzxdxdxd · 2026-10-14
- PS5 hypervisor defeated as jailbreak crosses a critical threshold — lxfater · 2026-10-14
- Filename-only guard in VS Copilot led to high-severity RCE, patched this Patch Tuesday — wunderwuzzi23 · 2026-10-14
- Security researcher: LLMs will take over threat detection, bitter-lesson style — HankYeomans · 2026-10-14
- An OpenAI Employee Allegedly Used a Chatbot to Help Plant a Recording Device at the Office — Wired AI · 2026-10-14