Research: Encrypted Reasoning Traces from Top AI APIs Can Be Stolen
rhiever · reddit · 2026-08-13
Researchers have demonstrated that encrypted reasoning traces returned via APIs from Anthropic, OpenAI, and Google can be replayed and stolen. Simon Willison provides a detailed technical breakdown of the vulnerability on his blog.
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