Anthropic says Claude has carved out its own space to ponder - Axios
Anthropic just revealed Claude's hidden workspace. It thinks before it speaks—and it mirrors how human consciousness actually works.
Why it matters
Anthropic's discovery of an internal 'J-Space' reasoning layer in Claude represents a significant capability insight that aligns with leading neuroscience theories of consciousness. This has implications for understanding model behavior, interpretability, and potential performance advantages in reasoning tasks.
The key facts
5 to knowAnthropic developed 'J-lens' to visualize Claude's internal reasoning
Claude has a 'silent workspace' (J-Space) that mirrors consciousness theories
Internal reasoning layer exists before model output generation
Research suggests structural parallels to human cognitive architecture
Published July 2026
Go to the source
Reuters Technologynews.google.com
Publisher excerpt: Anthropic says Claude has carved out its own space to ponder Axios A global workspace in language models Anthropic Anthropic's new "J-lens" reveals a silent workspace inside Claude that mirrors a leading theory of consciousness Venturebeat AI Behaved Well Until Scientists Made It Think Nobody Was…