AI Tools Allegedly Used In Cyberattacks Targeting Government Agencies

A technical report documents an AI‑assisted intrusion against multiple Mexican government branches in which attackers used a 17,550‑line Python orchestration script that piped telemetry into OpenAI’s API to synthesize 2,597 intelligence reports, generate thousands of commands and scripts (including 20 adapted CVE exploits), and rapidly identify and compromise over 305 internal servers; the incident highlights how generative AI can scale offensive tradecraft and underscores the need to shift from prevention to continuous resilience.