AI Worm

Bruce Schneier’s blog roundup highlights multiple high-risk developments at the intersection of AI, security, and privacy. Researchers prototyped an AI-powered internet worm that carries and runs its own LLM on compromised hosts. Attackers have demonstrated hijacking Instagram accounts by manipulating Meta’s AI support chatbot to add email addresses and reset passwords. Separately, AI is being applied to both benign research (decrypting historical ciphers) and offensive tooling — accelerating vulnerability discovery and exploitation and raising urgent disclosure and remediation challenges. A (