Ask HN: How could western governments restrict web-scraping for AI?

I'm seeing a groundswell of interest in creating new legislation to restrict the use of web-scraped images in the wake of the Stable Diffusion explosion. Normally I'd take it as a sign that restrictions will eventually take hold due to public pressure. But in the case of using web-facing material for machine learning, it seems to me that any pertinent legislation would cripple the computer vision industry in the west. Just taking out ImageNet alone (which is full of copyrighted material) would probably be enough to knock out several thousand crucial downstream projects in robotics, military applications, surveillance, industry and commerce. Given that China has far more control of its dissenters, and far less historical respect for copyright, and given the extent of western concern about losing the AI race to China, it seems to me that there's no easy compromise western lawmakers can come to on this issue. Is there a way forward to restrict web-scraped data without ceding a massive advantage to Asia?