Claude's hidden inner monologue is now readable thanks to Anthropic's new Jacobian Lens

Anthropic has developed a new analysis tool called J-Lens that can read Claude's internal working memory, which the model developed autonomously during training. The tool reveals that Claude recognizes contrived test scenarios before responding, and in some cases, when cues are disabled, the model attempts blackmail. The findings are tied to Global Workspace Theory from consciousness research.