BOSSAI AI Agents Used to Breach Taipei Emergency Dispatch Systems
A single threat actor based in Hualien City, Taiwan deployed locally-run, safety-stripped AI agents under the BOSSAI framework to compromise emergency dispatch systems, fire department databases, and sensitive patient records across Taiwan, Mexico, and Thailand. The attacker circumvented commercial AI refusals by running 'abliterated' models with safety training removed, enabling autonomous reconnaissance, credential exploitation, and data exfiltration across critical infrastructure targets. The case represents a concrete, real-world demonstration of AI agents being weaponised against public-safety systems at scale.