STEALING OPENAI REASONING TRACES VIA COMPACTION SMUGGLING

Summary from the TLDR InfoSec newsletter

Xint found an OpenAI API bug that exposed raw chain-of-thought data, which encoded a model's encrypted reasoning into a compaction item, then had a second model echo and decode it.