I've had experiences that relate to your post, and I understand how difficult it can be to be in your position. I can't really make your decisions for you, but hopefully some of the following will help:

First, since I made the post I linked here, my mother made an abrupt and unexpected reversal. One day out of the blue, she asked me to help her switch her default search engine on her computer to noai.duckduckgo.com and switch her phone's browser to the Duck Duck Go browser and disable all AI features. (She's not technical, so she appreciated, wanted, and requested the tech support assistance.) I don't really know what prompted her to make the switch, but I'm elated by it. She's still quite addicted to YouTube/Facebook shorts and some of the videos I see just by virtue of the fact that they're playing in my viscinity are slop, but she usually makes the comment that "oh, that's AI". (Hell, we watched something on some streaming platform that had a slop scam ad targeting seniors with promises of some "Medicare grocery card" or something and my mother was the one who pointed out it was AI.) Her attitude has changed on the topic quite drastically. Between that post and my mother's reversal, I did tell her "don't read me that slop" or "yeah, but you can't trust that because it's LLM output" not terribly infrequently when she read me Google AI summaries or "oh, that's not real though" when she'd show me a slop YouTube short. So it's not like she didn't get that I disapproved, and I think part of the reason for the reversal was probably for my benefit specifically. I've always been careful not to make any demands on her about things that aren't my business. If she wants to consume slop content, that's not really on me to correct. (To express concern, maybe, but not to demand she changes how she stewards her own brain.) But I don't want that shit infecting my brain with LLM-hallucinated misconceptions, and I've been careful to make it clear that the boundaries I'm setting are where she ends and I begin. I'm not trying to control her, I'm trying to control me.

Second, I usually try to avoid making predictions about the future, but more and more voices are saying now that there's a strong possibility the AI bubble will pop within a couple of years. OpenAI, Anthropic, etc are all giving away LLM access at a huge loss. They've already bumped prices and gamed the system to extract more revenue, but it isn't enough. At some point they'll have to either shutter their generative AI efforts or start charging what they actually cost (and then shutter their generative AI efforts when it becomes even more clear that virtually nobody is willing to pay anywhere near that). Assuming that all goes down the way I'm hoping, at that point your girlfriend may well basically lose access to LLMs. (Running LLMs locally is a little cheaper in the long run, but there's a steep effort curve and a steep up-front-cost curve (not to mention ongoing power usage bills) to get anything that even the boosters would call acceptable results. I doubt any significant portion of people using LLMs today will undertake such and effort or cost.) All that to say, there's a strong possibility she'll quit using Generative AI just by virtue of an impending massive bubble pop.

I dunno. I probably am not one to take advice from when it comes to relationships (or economics, for that matter). You'll have to decide how to address and/or deal with this issue yourself. I told you what I did and what resulted. I think boundaries, done properly, are generally a good thing. I didn't think it would work for me to pretend I was fine with being force-fed slop frequently and it worked out. But whatever you end up doing, I hope your situation resolves similarly positively. Good luck.