Machine Heart

Aug 12, 2026 · Information Security

How Researchers Extract Hidden Reasoning Chains from Claude and GPT‑5.6

A new security paper demonstrates that design flaws in Claude, GPT‑5.6 and other leading LLM APIs allow attackers to steal encrypted reasoning blocks, replay them in weaker compatible models, and reconstruct most of the hidden thought process, exposing privacy and safety risks.