Nadella Slams Claude Fable: "Too Editorially Controlled

Microsoft CEO Satya Nadella criticizes Anthropic's Claude Fable, stating the AI refuses too many requests and is overly restrictive. He believes its frequent refusals limit its usefulness, questioning if such heavy safeguards make sense for a creative AI tool. Nadella made these remarks during an internal meeting with engineers working on Microsoft's Copilot AI software. He suggests that Fable is too "editorially controlled." Anthropic introduced Claude Fable with stricter safety measures, which can automatically switch users to an older model for sensitive topics. This includes subjects like offensive cybersecurity, biology, and cancer research. Anthropic acknowledges these safeguards can lead to false positives and is working to reduce them. What's interesting is that Nadella's comments come as Microsoft focuses on cheaper AI models and reducing dependence on a few frontier AI companies. This matters because it highlights an ongoing debate about safety versus utility in AI development and its impact on user experience.