Re: First to legislate?!
Came here to write the same thing.
The best I can come up with is that the EU is probably a sufficiently large block that getting decent regulation out the door first will shape what the tech companies will do with their AI offerings and grasp the narrative early, before other countries that have more lax rules about what can happen with personal data shape it first.
Therefore it feels like it's not about what regulation the US puts in place after the EU sets its own regulation up, but to get the places working on various AIs to play ball appropriately.
The thing that's most interesting to me is how language affects availability of AI products. This isn't a situation like GDPR where an EU citizen is an EU citizen, this is a situation where an AI is probably going to be more fluent in English or Chinese than in French, German, Spanish, Portuguese etc etc. Now that the EU doesn't have a major (Sorry, Ireland) country where the official language is English, it's possible that non-english speakers will have been somewhat insulated from the majority of raw text data grabs used to train the LLMs.
tl;dr: EU is a big enough block that companies will probably think its worth it to work within reasonable EU regulations, but that depends on whether the e.g. Swedish market is big enough for an AI worth developing in that language in the first place.