[verdverm]

ChatGPT does not access websites at this point. You want to know if the public data set curators do, like CommonCrawl.

I would assume OP cares about their site being used for training more generally, like with LLAMA and everything else that will come out moving forward?

[drakonka]

Is that definitely the case? When I tried to ask it about the SVB incident, it first claimed I was mistaken and that nothing out of the ordinary was happening. When I persisted, it said it checked news sites to find more information since its latest data was from 2021 (though it claimed _not_ to find anything signaling SVB was in trouble, so I am not sure if it "lied" or just wasn't effective). When I linked to a specific news article describing the issue, ChatGPT responded in a way that made it clear it accessed the article and parsed its content.