AI’s hidden workspace mirrors how human consciousness works, says study
Researchers discovered a silent internal zone in Anthropic’s Claude models that behaves like a spotlight of consciousness, offering new insights into AI interpretability and safety.