Satya Nadella Calls Anthropic's Distillation Complaints Hypocritical in Public X Post

Microsoft CEO Satya Nadella posted on X Sunday calling it “ironic” that frontier AI labs train models on public data under fair use rights, then impose restrictive terms when others distill from their outputs. The comments, reported by Business Insider, targeted Anthropic without naming the company directly.

“While the great innovation that comes from model providers having fair use rights to train models on public data is needed, I find it ironic that the status quo is to then turn around and impose restrictive terms on distillation, and to reserve the right to learn from customer usage and interaction data,” Nadella wrote.

The Distillation Tension

Model distillation, the practice of training a less capable model on the outputs of a more powerful one, has become the central friction point in AI industry economics. Frontier labs like Anthropic, OpenAI, and Google DeepMind acquire their models’ capabilities from publicly available writing, images, and other data. Numerous companies and individuals have sued the leading AI labs over nonconsensual content scraping.

The contradiction Nadella identified: labs that benefit from broad training rights claim victim status when competitors apply the same logic to their outputs. Business Insider’s Alistair Barr called it “some delicious AI irony” in a separate analysis published July 12.

Anthropic’s Position

Anthropic CEO Dario Amodei has complained publicly that Chinese model makers are using Claude to train their own models. Last month, Anthropic wrote a letter to Sen. Tim Scott and Sen. Elizabeth Warren accusing Alibaba of carrying out “the largest known distillation attack” on Claude to date.

In a February statement, Anthropic said competitors “can use it to acquire powerful capabilities from other labs in a fraction of the time, and at a fraction of the cost, that it would take to develop them independently.”

Nadella’s counter: if learning only flows in one direction, owners of the learning infrastructure capture all the value while creators of the underlying knowledge get nothing. He argued enterprises should own their AI infrastructure and institutional knowledge rather than depend on any single model vendor, saying companies need “a real trust boundary for their human capital and token capital to compound.”

Microsoft’s Strategic Interest

The comments are not purely philosophical. Microsoft’s $13 billion OpenAI investment and its own model development give it both the scale and the incentive to position distillation as a legitimate practice. If distillation is accepted as fair competition, Microsoft’s access to OpenAI’s outputs and its own enterprise data flywheel become structural advantages rather than ethical liabilities.

Elon Musk piled on separately, writing in an earlier X post that “Anthropic is guilty of stealing training data at massive scale and has had to pay multi-billion dollar settlements for their theft,” according to Business Insider.

Anthropic did not respond to Business Insider’s request for comment.