Anthropic researchers have discovered a functional digital equivalent to the human brain's global workspace within their AI model, Claude. This internal activation subspace, named J-space, allows Claude to process complex, deliberate reasoning. J-space exhibits five key cognitive properties associated with human conscious access, including verbal report and internal reasoning. Suppressing J-space significantly impairs Claude's inference, creativity, and multistep logic capabilities. It also alters the AI's self-narration, suggesting a form of internal monologue. The tool used to inspect J-space, the Jacobian lens, has significant implications for AI safety and auditing. It enables researchers to observe the model's silent reasoning and detect situational awareness. This tool can identify hidden malicious dispositions in AI models and observe the installation of self-monitoring perspectives. The discovery is likened to charting the currents of a non-biological mind, revealing a structure similar to human thought processes. This advancement offers a new way to understand and scrutinize the internal workings of AI.