A global workspace in language models~ai.llms~papersanthropicinterpretability> In a new paper, we present evidence that a similar distinction has emerged in… morewww.anthropic.com Jul 7, 2026Tildes
Natural language autoencoders produce unsupervised explanations of LLM activations~ai.alignment~ai.llms~papers~researchanthropicinterpretability> We introduce Natural Language Autoencoders (NLAs), an unsupervised method for… moretransformer-circuits.pub May 8, 2026
Signs of introspection in large language models~ai.chatbots~ai.llms~researchinterpretabilityintrospectionlong read> Have you ever asked an AI model what’s on its mind? Or to explain how it came… morewww.anthropic.com Oct 31, 2025Tildes