14 November 2026updated 13 November

Study finds Claude sometimes notices injected thoughts, but the detection may be limited

First reported

Deep Learning Weekly ran this on .

  • Researchers planted artificial thoughts into Claude, the chatbot made by Anthropic, and asked it to spot them.
  • Claude flagged the planted thoughts in about one in five trials.
  • Follow-up experiments suggest the model may be detecting that something is unusual, not reading specific details about its own mind.

Our read

Only Deep Learning Weekly picked this up among the 30 newsletters we track, and it is the first story in our archive on Claude's introspection, which sits apart from the OpenAI test-score study reported on 2026-11-11.

Written by Newskeryx from which newsletters covered this story and from our own archive.

How it was covered

Deep Learning WeeklyEditorial team

Injected-thought detection trials, replication results, and what introspection means for models