Nadella Slams AI Model Makers Over Data Training Practices

Microsoft CEO Satya Nadella criticized AI model makers like Anthropic for their training practices. He stated that it's hypocritical for them to complain about "distillation" when they themselves train models on public data. Distillation is when a less powerful model learns from a stronger one. Nadella posted on X that if learning only flows one way, the owners of the learning infrastructure make all the money, leaving out the creators of the knowledge. He implied that companies relying on leading models are giving away their proprietary data. Nadella believes companies should own their AI infrastructure and knowledge. This matters because it impacts how businesses protect their data and intellectual property in the age of AI.