AI aggregating news in its current state is very dangerous the core problem is analyzing the reputability of a claim (for example, grok regularly calls me solana ceo on its summaries) it can look at likes, but likes can be gamed it can look at views, but those can be gamed it can look at the number of different accounts saying something similar, but that can be gamed it can combine the above with some first principles thinking, but for "news" events that can't actually be confirmed/denied, it will get skewed by sentiment it can try, in some way, to verify authority, i.e this person is the ceo of such and such protocol and they might have better information, but then i) you need to submit a bunch of info on your x account, and then ii) the authorities can also lie! best is probably to skew it to be extra conservative until it has at least an 80% confidence threshold based on a mix of public sentiment, authoritative sources, wide internet search, and perhaps even prediction markets but the current state is actively misleading