Nadella: Anthropic's AI Model Restrictions "Illogical

Microsoft CEO Satya Nadella calls Anthropic's model restrictions illogical. He publicly criticized AI model providers, including Anthropic, for what he sees as a contradiction. Here's the thing: these companies claim fair-use rights for publicly available data to train their models. Then, they restrict customers from using those same models for distillation and retain rights over customer interaction data. Model distillation involves training a smaller model using a larger one. What's interesting is the timing. Anthropic faced U.S. export controls in June 2026, which suspended advanced capabilities for two models, Claude Fable 5 and Mythos 5. While global access to Claude Fable 5 was restored, Mythos 5 remains restricted to approved organizations in the U.S. Mythos 5 is Anthropic’s most capable offering in that product line. Nadella warns about economic concentration if a few AI models become essential global infrastructure, subject to export controls and licensing. This gives model developers significant leverage. Microsoft is promoting its own in-house models as alternatives. The bottom line: this signals that major cloud providers are working to reduce customer reliance on single AI model providers. This could mean competitive pressure on Anthropic’s pricing and terms.