Re:Stealing Reasoning Traces from Proprietary LLM APIs
by orbital-decay on Tuesday August 11, 2026 @12:20PM
(#49260603)
Not necessarily. There's a million ways to jailbreak any current model to show the trace and bypass all guardrails, or hijack and modify it. It's just one of them.
[ Parent
| on HN ]
There are no comments on this comment yet.