Bans in general don't have to be and rarely will be completely enforceable in all cases. But a ban with significant enough consequences would mean most businesses wouldn't think about trying them at some point, just to avoid the risk.
Weights are fungible.
I fine-tune an open weigh model and call it legit. Good luck for authorities to prove where the base model was from, or to prove a Tor connection a few months ago was fetching suspicious bytes.
Not to mention that as enterprise demand for newly "illegal" LLMs dries up, so will the incentive for Chinese labs to shovel money into building them. I doubt Chinese labs are getting rich off providing inference to American business, but loosing them would be a permanent dent, along with rendering moot the perhaps more nefarious incentives Chinese labs have to release model weights in the first place.
BTW - When was the last time you saw mentions of DeCSS?