Anthropic has developed a new tool called the "J-lens" that appears to reveal a hidden workspace within its language model Claude. This workspace, also referred to as "J-space", seems to mirror a leading theory of consciousness and may indicate that Claude is capable of silent reasoning or holding unspoken thoughts. The nature and implications of this discovery are still being explored, with some outlets describing it as a "consciousness-like workspace" and others noting the unusual insights it provides into how AI models process information.