There is a bit of an arms race between id verification systems and users bypassing them when AI gen. Which is really just ai generated images vs. AI generated image detection.

In practice, nothing will stop it, the tooling will gradually get better at detecting prior fakes and banning those users while the newer fakes will go undetected for longer.

Putting up the requirement satisfies their CYA requirements here. The race between AI fraud vs. detection is something they can just ignore and let happen on its own.