Microsoft CEO Satya Nadella has publicly criticized AI developer Anthropic for the "overly restrictive safeguards" imposed on its latest model, Claude Fable 5. Speaking to employees, Nadella reportedly expressed frustration, stating that such tight controls on a creative tool "doesn't make sense."
Nadella specifically targeted Anthropic's policy against "model distillation," where the AI reportedly switches to an older version if a user's query could aid in developing a competing AI. He questioned the logic of a creation tool being so "editorially controlled," highlighting instances where Fable 5 refused seemingly innocuous prompts.
Why Anthropic Imposes Restrictions
Anthropic's cautious approach stems from its desire to prevent competitors from replicating Claude Fable's advanced capabilities without equivalent investment in research and development. The company views such "distillation" as effectively copying years of expensive AI innovation.
An Unexpected Critique Amid Partnership
The public critique is notable given the close business relationship between Microsoft and Anthropic. Microsoft previously invested $5 billion in Anthropic, and Anthropic, in turn, committed $30 billion to utilize Microsoft's Azure cloud platform for training and running its AI models. Microsoft also relies on Anthropic's Claude Code.
Beyond this specific criticism, Nadella reiterated his broader vision for enterprise AI. He advocates for businesses to build their own cost-effective AI models using internal data, emphasizing the importance of owning their AI infrastructure to avoid dependency on a single AI provider.