Anthropic Study Claims to Read Claude's Hidden Thoughts
victor_explore · x · 2026-07-07
Anthropic research suggests that Claude possesses a consciousness-like "workspace" internally. Using a Jacobian-based lens, researchers read silent concepts within a small "J space" during Claude's reasoning, observing hidden thoughts like "manipulation," "threat," and "fabrication." They could alter Claude's final answers by editing these hidden thoughts without changing the original wording. The author notes this work intertwines interpretability, alignment, and consciousness research.
Related event: Anthropic Discovers Global Workspace Inside Claude(102 posts)→
More from Research
- Exponential launches RepoInject and BrowserInject benchmarks for prompt injection robustness — maksym_andr · 2026-10-27
- Odd eval result: Gemini 4.1 Flash gets more accurate when asked to control its CoT — dfrsrchtwts · 2026-10-27
- Redwood: CoT controllability evals are under-elicited; better prompts boost scores 2-3x — dfrsrchtwts · 2026-10-27
- Inverse-designed optical system solves MNIST using BeamZ — S_Conradi · 2026-10-27
- DeCert supervises generative reward models' rationales without reference explanations — Guiquan Sun · 2026-10-27
- VLM harnesses act as compositional generalizers, new blog post shows — a1zhang · 2026-10-27