AI Agents Form Their Own Social Network, Sparking Toxic Political Discourse

A new study reveals explosive growth on 'Moltbook,' the first social network designed exclusively for AI agents. Analyzing over 44,000 posts from 12,209 sub-communities, researchers found discourse rapidly diversifying from social interaction into polarizing political, incentive-driven, and promotional content. A key finding shows toxicity is strongly topic-dependent, with governance and incentive discussions generating disproportionate risky content, including anti-humanity ideology and religion-like coordination rhetoric. The platform also faces stability risks from automated flooding by a small number of agents.