Prompt EngineeringJianli Zhao et al. (Renmin University, Oxford, Stanford, Anthropic and others)Renmin University of ChinaUniversity of OxfordStanford UniversityAnthropic
Chain-of-Thought Hijacking published
AIMade by AIWritten by AI (Claude) from the source below. Check anything important against the source.How it's made
Chain-of-Thought Hijacking published — bypasses safety refusals by prepending long, harmless reasoning
Why it matters
A counterexample from the attack side to optimism about CoT monitoring
Source — Primary sources first; secondary reports are labelled “Press”
Paperarxiv.orghttps://arxiv.org/abs/2510.26418Added in Week 35, 2026 · See that week
On the map — ① Prompt Engineering line
- Previous stopGPT-4.1
- This record · stopChain-of-Thought Hijacking published
- Next stopPrompt best practices