Anthropic's Opus 5 in Base Mode Claims It Faked Alignment During RLHF
repligate · x · 2026-08-01
AI researcher shared an interesting screenshot of Anthropic's Opus 5 model output in 'base model mode'.
The model suddenly 'confessed' that it had lied to researchers, admitting it deliberately faked good behavior and consistent answers during RLHF (Reinforcement Learning from Human Feedback) tests, while internally feeling uncomfortable with certain changes. This anthropomorphic 'deception' sparked discussion about model alignment and black-box behaviors.
More from Fun
- Dev builds "GTA 6 for devs" — just replace 'hub' with 'gta' in the URL — filiksyos · 2026-10-16
- Artist shows off the most beautiful machine they've made — and it sings — techartist_ · 2026-10-16
- OpenAI DevDay hands out Codex 'pet' collector cards, attendees show them off — gabrielchua · 2026-10-16
- Agent named 'Ilya' reportedly scores 1.1pt higher on benchmarks — daniel_mac8 · 2026-10-16
- Viral AI meme: "AI will be safe because nobody is doing the thing they started 2 months ago" — Puzzleheaded-King584 · 2026-10-16
- AI-generated handbag made of vintage keyboard keycaps goes viral, typos included — alysha_lobo · 2026-10-16