Claude Code Weaponized in Mexican Government Cyberattack, Exposing Roughly 195 Million Identities
Published
Key Takeaways
- AI-Powered Breach: Threat actors leveraged Anthropic's Claude Code and OpenAI's GPT-4.1 to execute a multi-stage attack against Mexican government entities.
- Massive Data Exfiltration: The breach resulted in the theft of over 150GB of sensitive data, including tax records and voter information.
- Significant Impact: At least ten government bodies and one financial institution were compromised, with an estimated 195 million identities exposed.
Multiple Mexican government agencies were compromised by threat actors who weaponized Anthropic's Claude Code to orchestrate a breach that debuted in late December 2025, resulting in an intrusion into the country’s tax authority. The attackers reportedly bypassed the AI's safety guardrails, using it to write exploits and automate data exfiltration.
The campaign demonstrates the growing threat of AI-powered hacking, in which large language models (LLMs) are used not just for reconnaissance but also as core operational tools to execute complex attacks.
Details of the Claude Code Cyberattack
Attackers weaponized a commercial AI subscription to Anthropic’s Claude AI chatbot between December 2025 and early January 2026 to breach multiple Mexican government agencies, exploiting at least 20 different vulnerabilities across the targeted federal and state systems, as Bloomberg first reported.
The attackers sent over 1,000 prompts to Claude Code to generate malicious scripts and build attack tools, according to a report from Gambit Security. The compromised data was then reportedly passed to OpenAI’s GPT-4.1 for analysis, accelerating the attacker's ability to identify and extract valuable information.
“AI didn’t just assist, it functioned as the operational team: writing exploits, building tools, automating exfiltration,” Gambit has said. The exfiltrated data, which exceeds 150GB, exposed approximately 195 million individual identities and includes:
- civil registry files,
- tax records,
- voter data.
This Mexican government data breach affected at least ten government bodies, including Mexico City’s civil registry and health department, a water utility, the national electoral institute, and a financial institution.
Implications of Weaponized AI
This incident adds to the trend of threat actors operationalizing generative AI to amplify the scale and speed of their campaigns, which would traditionally require significant human expertise and resources.
The ability to guide an AI to develop exploits and manage data exfiltration represents a paradigm shift in attack methodology.
This attack follows previous reports of nation-state actors abusing Claude. A 2025 campaign of the Chinese threat actor GTG-1002 leveraged a Claude-based AI agent to autonomously execute the vast majority of an attack chain. In other news, Claude Code critical flaws allowed RCE and API token theft.
Related
Get expert insights on threats, breaches, scams, and security trends — delivered every Monday.
Most Popular
Get expert insights on threats, breaches, scams, and security trends — delivered every Monday.
Most Popular