American companies are spending enormous sums to develop high-performing AI models. Distillation attacks are attempting to maliciously extract them — and nobody is doing much to stop it.
American companies are spending enormous sums to develop high-performing AI models. Distillation attacks are attempting to maliciously extract them — and nobody is doing much to stop it.
Actually, Chinese are doing a whole lot more innovation than American “AI brains,” or at least innovation that we know about. Architectures are getting more and more efficient, instead of US Big Tech’s “the same, but bigger, and capture regulators” ethos.
Not that the Chinese labs are saints. They’re 100% distilling US labs data. It’s somewhat measurable:
https://eqbench.com/creative_writing.html
They’re almost certainly using unspecified Chinese govt data too, or at least sharing data between them, given the common quirks and behavior across models and the efficiency for their size. Not to speak of political “gaps” (which US models certainly have too).