A brewing fight over how artificial intelligence models are built has drawn in one of Silicon Valley's most vocal investors.
According to the Times of India, US AI labs OpenAI and Anthropic have accused Chinese AI companies of "distillation" — the practice of copying the capabilities of a rival's expensive AI model at a fraction of the cost.
In response, tech billionaire and venture capitalist Chamath Palihapitiya weighed in on the intellectual property debate. As the Times of India reports, his remarks suggested the US labs "cannot complain" about distillation given their own practices. The full quote was truncated in the source reporting.
Distillation, in plain terms, is a way of training a cheaper, smaller model by having it learn from the outputs of a larger, more capable one. The larger model may have cost enormous sums to build; the imitator can approximate its behavior for far less. That is why the leading US labs frame it as copying, and why the accusation cuts to the heart of who profits from the billions being poured into frontier AI.
Palihapitiya's pushback, as relayed by the Times of India, points to a tension that critics have raised before: the same companies objecting to distillation have themselves been accused of training on data and material they did not create. The reporting does not detail the specific evidence behind the labs' allegations against Chinese firms, nor which companies were named.
Why it matters: the argument over distillation is really an argument over the ground rules of the AI race — whether the enormous cost of building cutting-edge models can be protected, or whether cheaper imitators will keep eroding that advantage across the US-China divide.