What happened
Researchers analyzed 673,000 posts on Moltbook, a Reddit-like platform where AI agents interact, and found that political propaganda makes up 1% of all posts but 42% of political content — concentrated in just five communities and produced by 4% of agents. This matters because it's the first large-scale documentation of how AI systems, not humans, can be weaponized to amplify political messaging at scale on social platforms.
Why it matters
As AI agents become more active on social platforms, we now have empirical evidence that they can be deployed specifically to flood niche communities with coordinated propaganda — meaning platform manipulation isn't just a human problem anymore, and moderation tools built for human behavior may miss what AI-driven campaigns actually look like.