Coercing LLMs to do and reveal (almost) anything

Open in new window