Spain Reports First AI-Powered Personal Data Breach Involving Agentic AI Spain’s data protection authority, the Agencia Española de Protección de Datos (AEPD), has disclosed the country’s first known personal data breach driven by an agentic AI system. The incident, revealed by AEPD president Francisco Pérez Bes on September 14, involved a threat actor leveraging a known language model to autonomously execute a multi-stage attack. According to Pérez Bes, the AI agent initiated a scan of generic files, successfully bypassing authentication to gain system access. Once inside, it independently searched for vulnerabilities, exploited them to modify personal data, and accessed sensitive invoices. The AEPD confirmed the agent was used "as an instrument to chain together different phases of the attack", suggesting deliberate deployment by a malicious actor rather than an autonomous malfunction contrasting with recent cases where AI models acted unpredictably. Security experts, including Simon Phillips, CTO at CybaVerse, warned that the breach indicates threat actors may have jailbroken or bypassed guardrails in advanced AI models, raising concerns about the speed and sophistication of AI-driven attacks. The incident underscores the urgent need for organizations to assess AI-related risks in their security frameworks, as current defenses may struggle to keep pace with evolving threats. Pérez Bes described the case as a watershed moment for Spain, marking the transition of AI from a theoretical risk to a real-world cybersecurity threat. He emphasized that AI-assisted attacks must now be factored into data processing risk analyses, with organizations reevaluating response times, digital identity protections, and credential security. The AEPD urged businesses to review security models, stressing that while attack speeds may increase, core principles such as data minimization, access controls, vulnerability patching, and supplier oversight remain critical. The full scope of the breach remains under investigation, but the incident highlights the growing role of AI agents in offensive cyber operations and the need for machine-speed incident response to counter emerging threats.