I'm genuinely curious about your view on this. Ziphu AI release pretty capable open weights models. As long as we don't feed information that is confidential or become overy reliant on the tech that's on someone else's computer - what are the other riaks that you see in using Chinese models?
That doesn't make any sense. Model weights are literally just numbers you multiply by. To imply that it's any way even remotely comparable to the xzutils build process is...
Well, I would be very technically impressed if someone managed to achieve any form of code execution, especially given unknown levels of quantisation post-release whereas xzutils was interesting mostly due to obfuscation.
Steering a single model towards that seems, again, technically challenging, especially if it needs to be obfuscated enough to pass code review (you are doing that for LLM code right?) and especially if targeted towards specific fields of use (e.g. critical infrastructure). I still don't see how the threat model could make sense here.
