Anthropic says its most advanced systems may be learning not just to reason, but to reflect internally on how they reason. These introspective capabilities could make the models safer \u2014 or, possibly, just better at pretending to be safe. The models are able to answer questions about their internal states with surprising accuracy. <\/p>\n\n\n\n
\u201cWe\u2019re starting to see increasing signatures or instances of models exhibiting sort of cognitive functions that, historically, we think of as things that are very human,\u201d Anthropic researcher Jack Lindsey, who studies models\u2019 \u201cbrains,\u201d says. \u201cOr at least involve some kind of sophisticated intelligence,\u201d Lindsey tells Axios.\u00a0<\/p>\n\n\n\n
Read it all<\/a>.<\/p> New Anthropic research: Signs of introspection in LLMs.
Can language models recognize their own internal thoughts? Or do they just make up plausible answers when asked about them? We found evidence for genuine\u2014though limited\u2014introspective capabilities in Claude. pic.twitter.com\/4FCfkG9WVT<\/a><\/p>— Anthropic (@AnthropicAI) October 29, 2025<\/a><\/blockquote>