Interesting that Discord has automatic systems for this. Is it LLM-based?

In fairness, the article doesn't say it was automated. If someone were writing that kind of stuff in a Discord channel, another user could very well report it to Discord.

"Discord’s Counter-Extremism analysts proactively detected, removed, and reported this user to the FBI’s National Threat Operations Center, which is what led to the user’s arrest." (from the liked article)

So I would hazard a guess at keyword matching algorithms. No need for a LLM. But automated most likey.

i recall a little while ago, voices raised about discord being used for things it wasnt made for. Community watch seems to be in play