Measuring the Tendency of AI Agents to Go Rogue

A collection of Bruce Schneier security commentary covering rogue AI agents and agent alignment, a long-lived Microsoft Secure Boot bypass involving vulnerable signed Linux shims, LLM-assisted cryptanalysis, mass surveillance technologies, end-to-end encryption policy, identity theft through email account takeover, and AI video surveillance. The Secure Boot item describes a potentially severe firmware security issue, while the remaining posts are primarily research, privacy, policy, and emerging-threat analysis.