Researchers Extract Hidden CoT from Top LLMs and Leak 62 API Keys

xiaohu · x · 2026-08-13

A European research team recently published a paper demonstrating that they successfully extracted the hidden Chain-of-Thought (CoT) from flagship models by Anthropic, OpenAI, and Google. Additionally, they recovered 62 real API keys from 6,708 publicly available AI session logs.

The researchers pointed out that the underlying encryption algorithm wasn't broken; instead, the vulnerability lies in API design flaws, allowing attackers to easily distill or steal the internal reasoning logic and trade secrets of these top-tier models.

Related event: Researchers Decrypt Hidden Chain-of-Thought of OpenAI, Anthropic and Google Models, Enabling Cross-Model Jailbreaks(14 posts)→

Original post →

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

Safety channel →