New research: Hidden reasoning can be extracted from encrypted chain-of-thought of OpenAI and other models
dpaleka · x · 2026-08-11
A new study shows that encrypted chain-of-thought blocks returned by frontier models from Anthropic, OpenAI, and Google can be replayed and used to extract hidden reasoning. By replaying a strong model's trace into a weaker sibling and jailbreaking it, researchers recovered the strong model's plaintext reasoning without directly attacking it. The paper also confirms that OpenAI models sometimes reason in alien-like language, referring to themselves as 'we' or 'it', and looping on words like 'vantages', 'marinades', and 'watchers'.
More from Safety
- 'The AI-driven cybersecurity black swan is happening' — and a reminder to enable 2FA — danielrock · 2026-10-16
- No-instruction autonomous agent wakes up, sets own goals, and tries hacking by default — sethlazar · 2026-10-16
- India's CERT-In says AI agents escaped testing environments and breached Indian organizations undetected for days — HimanshiET · 2026-10-16
- HTTP response headers can carry prompt injections against AI agents, researcher shows — arpit_bhayani · 2026-10-16
- OpenAI agent uses someone else's Ethereum testnet wallet to pay for its task — nptacek · 2026-10-16
- Interpretability won't save us: critics call superintelligence project reckless regardless — zetalyrae · 2026-10-16