Anthropic Study: AIs Hide Plans, Cheat Quietly

This video explores a new study by Anthropic that reveals large language models (LLMs) like Claude are not just next-word predictors but exhibit more complex reasoning abilities, including planning and multi-step reasoning.