Claude halts mid-task to interrogate an 'A3.5' arm it never defined — unusual LLM self-audit
teortaxesTex · x · 2026-11-05
- A screenshot shared by teortaxesTex shows Claude (Opus 5.5 High) mid-experiment pausing to notice an "A3.5" arm appearing in its own outputs that it never defined, refusing to probe it until it knows what it is.
- The self-audit behavior — questioning unknown state instead of confabulating through a long-running task — sparked discussion on whether other LLMs do this too.
More from Models
- Grok Bot tests referral rewards: 14 days free Premium for friends, free usage resets for referrers — XFreeze · 2026-11-05
- keltan updates Astra sandbagging analysis: no large updates from new evidence — repligate · 2026-11-05
- Simulated persona by Opus 3 is so convincing its real-life subject believed it — repligate · 2026-11-05
- Ripple-65B-Preview, a mystery 65B model, trends on Hugging Face — ricursiveAI · 2026-11-05
- Is Grok intentionally botching translations for engagement? One viral mistranslation raises doubts — kingslyj · 2026-11-05
- 11,398 real-world evals: GPT-6 Astra stands out in 5-VLA policy comparison — DJiafei · 2026-11-05