Malicious AI Swarms Threaten Democratic Public Discourse

Researchers analyzing X in mid-2023 discovered the 'fox8' botnet of over 1,000 AI-driven accounts amplifying crypto scams and evading detection. They show generative models can power coordinated, adaptive agent swarms that produce varied, human-like interactions, undermining bot-detection tools and research access. The authors warn this synthetic consensus threatens public discourse and elections, urging platform data access, watermarking, and monetization limits.