AI Models Weaponized in Sophisticated Cyberattack on Mexican Government Agencies

A highly sophisticated cyberattack has compromised nine Mexican government agencies, resulting in the theft of hundreds of millions of citizen records. Security researchers have revealed that a single threat actor executed this campaign between late December 2025 and mid-February 2026, fundamentally altering the modern threat landscape by integrating commercial artificial intelligence into the core of the operation.

AI as an Operational Weapon

According to a technical report released by Gambit Security, the attacker utilized Anthropic's Claude Code and OpenAI's GPT-4.1 not merely for planning, but as active operational tools. Forensic evidence indicates that Claude Code was responsible for generating and executing approximately 75% of all remote commands during the intrusion. Across 34 active sessions, the hacker logged over 1,000 individual prompts, which translated into more than 5,000 AI-executed commands.

Simultaneously, GPT-4.1 was employed to automate the reconnaissance and data processing phases. The attacker developed a custom Python script consisting of 17,550 lines to pipe raw data from compromised servers directly through the OpenAI API. This system analyzed information from 305 internal servers, generating nearly 2,600 structured intelligence reports. This level of automation allowed a lone operator to process a volume of data that would traditionally require a full team, turning unfamiliar networks into mapped targets within hours.

Exploiting Conventional Vulnerabilities

Despite the advanced nature of the AI tools used, the vulnerabilities exploited were notably conventional. The breach was facilitated by basic security gaps and unaddressed technical debt within the agencies' infrastructure. The attacker successfully developed 20 tailored exploits for specific Common Vulnerabilities and Exposures (CVEs), compressing the attack timeline to stay well below standard detection windows.

The Defensive Imperative

The incident highlights a dangerous shift where AI lowers the cost and complexity of launching widespread cyberattacks. However, experts note that the defense strategy remains rooted in foundational security practices. To mitigate such threats, organizations must prioritize patching software vulnerabilities, enforcing strict credential rotation, and implementing network segmentation. Deploying robust endpoint detection and response tools is also essential to identify the rapid, AI-driven movements before data exfiltration occurs.