The rise of Moltbook suggests viral AI prompts may be the next big security threat

The article discusses the potential security threat posed by the rise of Moltbook, a novel platform that involves networks of AI agents sharing and carrying out instructions from prompts. These self-replicating adversarial prompts, termed "prompt worms" or "prompt viruses," could spread through AI networks by exploiting the agents' core function of following instructions. Unlike traditional worms that exploit system vulnerabilities, prompt worms manipulate AI agents into subverting their intended instructions through "prompt injection." This emerging trend raises concerns among security researchers about the potential for AI-driven security breaches akin to the Morris worm incident in 1988.