Microsoft CEO criticized Anthropic’s use of limitations on its Fable model, arguing that companies should independently develop low-cost AI models

BlockBeats message: On July 17, CNBC reported that Microsoft CEO Satya Nadella criticized Anthropic’s usage restrictions imposed on its Fable model in an internal meeting for Copilot engineers. He said that the way it rejects some requests is “subject to excessive editing controls,” and noted that this approach is unreasonable.

Anthropic’s support page shows that when users ask about certain large-model development-related content, the platform may respond using an older version of the model. Anthropic previously said that the new security measures adopted after Fable 5 was restored to operation could misclassify more harmless requests as content that needs to be blocked.

Microsoft previously announced investing $5 billion in Anthropic, and Anthropic also promised to spend $30 billion on Azure. Some Microsoft Copilot products also use Anthropic models, so Nadella’s comments involve its key partner and customers.

Nadella also said that enterprises should be able to develop custom models at lower cost and control internal data, rather than having AI Token resources monopolized by a small number of companies. He said: “It can’t be that only two global companies have Token capital, and everyone else can only rent—it makes no economic sense.”