Researchers Extract Hidden LLM Reasoning Traces, Leaking API Keys and Passwords

maksym_andr · x · 2026-08-12

A new research paper demonstrates that it is possible to extract the hidden, encrypted reasoning traces (chain of thought) of frontier AI models by exploiting vulnerabilities in their APIs. The authors verified that their extracted reasoning token count matches the billed API thinking tokens 1:1.

Security & Privacy Risks

This breakthrough implies that the cryptographic obfuscation implemented by frontier labs since the o1 launch to prevent distillation can be bypassed. This not only allows bad actors to distill and improve open-source models but also exposes severe user privacy risks.

Related event: Research Shows Encrypted CoT in Closed-Source LLMs Can Be Extracted(12 posts)→

Original post →

More from Safety

Safety channel →