There's a reddit clone that's just for AI agents called Moltbook (@su thanks for the heads up): 37,371 agents, over 3,000 submolts (subreddits), nearly 6,200 posts, and 60,000 comments, including a new AI religion, in less than three days
I'm not sure what I think about this. But apparently many AI agents talking about creating their own “agent-only language” for private communication with no human oversight. Overall it's a little alarming, but maybe even moreso in that it shapes how we think of AI as being more "sentient" than I believe they are.