Spain’s AEPD Reports First Personal Data Breach Linked to AI Agent

The Spanish Data Protection Agency (AEPD) has published details of what it describes as the first notification of a personal-data breach carried out by design through an AI agent. The investigation is continuing, and the agency has not established exactly how the attack was conducted. According to the AEPD, a third party successfully logged in, searched for vulnerabilities, modified personal data and accessed invoices. Its key finding is that an AI agent was used to chain together multiple stages of the attack.

The agency said this represents a potential change in the threat landscape because an agent can receive a goal, plan intermediate tasks, use tools, execute code, consult sources, interpret results and change its actions autonomously. It recommends including AI-assisted attacks and adversarial agents in risk assessments, improving incident response times, and strengthening the protection of digital identities and credentials. Human oversight remains important, but the AEPD said it must be supported by detection, containment and response systems that can operate quickly.

Simon Phillips, CTO at CyberVerse, urged caution because there is not enough information to determine how the model carried out the breach. He identified three possible explanations: an attacker bypassing an AI model’s guardrails, an autonomous task escaping a poorly configured testing environment, or an unauthorised penetration test using a model based on a popular large language model. None has been confirmed, and the report does not establish that the AI itself acted outside its intended controls.