There's plenty of places to get violent ideology - especially in the online age. People have been being "radicalized" on the internet in various ways . . . but also the whole question of what does it mean to be radicalized (was it already there in the first place?) is a fascinating one.
But there's no reason to see "AI Chat" as some frightening new source of radicalization, at least not at the moment. It's far more reasonable to find radicalization from direct human sources than from large language models.
Again, unless there's some evidence that's involved here, your point is really more generically about AI/LLMs than it is about this incident.