Alleged Unauthorized Access to Anthropic's AI Model Claude: Data Breach of Mexican Government Systems

December 1, 2025

In December 2025 to January 2026, an unknown attacker reportedly gained unauthorized access to Anthropic's AI model, Claude. The attacker allegedly used the model to identify vulnerabilities and generate exploitation scripts, which were then utilized in a campaign against Mexican government systems. These actions are believed to have resulted in the theft of approximately 150 GB of sensitive data, including taxpayer, voter, government employee, and civil-registry information. This incident underscores the need for robust AI governance and harm prevention measures. For more information on how you can contribute to responsible AI discussions and help ensure safe and secure AI development with HISPI Project Cerebellum TAIM, visit JOIN US. Through your engagement, we can collaborate towards building a trustworthy digital future.

Alleged deployer

unknown-hacker

Alleged developer

anthropic

Alleged harmed parties

mexican-taxpayers, mexican-voters, mexican-government-employees, state-government-of-tamaulipas, state-government-of-michoacan, state-government-of-jalisco, monterrey-water-and-drainage-services, servicio-de-administracion-tributaria-(sat), instituto-nacional-electoral-(ine), direccion-general-del-registro-civil-de-la-ciudad-de-mexico-(dgrc)

AI governance case studies

For forensic AI governance failure analysis (TAIMScore™ case studies), browse Human Signal’s Failure Files™.

Data source

Incident data is from the AI Incident Database (AIID).

When citing the database as a whole, please use:

McGregor, S. (2021) Preventing Repeated Real World AI Failures by Cataloging Incidents: The AI Incident Database. In Proceedings of the Thirty-Third Annual Conference on Innovative Applications of Artificial Intelligence (IAAI-21). Virtual Conference.

Pre-print on arXiv · Database snapshots & citation guide

We use weekly snapshots of the AIID for stable reference. For the official suggested citation of a specific incident, use the “Cite this incident” link on each incident page.