Six Chinese AI firms accused of aggressively copying US frontier models
arstechnica.com
arstechnica.com
But Now it's "industrial-scale theft" requiring a joint NSA/FBI/CISA advisory.
The hypocrisy. Whatever we do is fair use. Whatever you do is a federal case.
Chinese AI firms in my opinion are much better, DeepSeek publishes weights, Qwen publishes weights, GLM is a godsend (imho). You can download them, audit them, run them locally all while paying lesser than American AI. The "shady" companies are the transparent ones. The "responsible" American companies are the ones suing to keep their models closed while selling access to the NSA
Towards the end of his piece yesterday, "The Case for Boycotting Generative AI" [0], Marcus also backs calls for co-operation with China, rather than antagonism and bickering. After all, tackling a claimed existential threat should take precedence over any other considerations. This seems reasonable to me.
[0] https://garymarcus.substack.com/p/the-case-for-boycotting-ge...