Loading page…
The preliminary notification describes an agent moving from access to vulnerability discovery and data changes. The case is still under investigation, but AEPD says companies should account for AI-driven attacks in their risk assessments.
Listen to this story
Spain’s data-protection authority is investigating a reported breach in which an autonomous AI agent allegedly used public files to access a company system, find and exploit a vulnerability, alter personal data, and retrieve invoices. The notification is preliminary: AEPD has not established that the underlying model or its provider was compromised or malicious.
AEPD says the agent reportedly chained reconnaissance, exploitation, data modification, and invoice access.
The authority has only the affected organization’s notification; the investigation remains incomplete.
The incident does not show that the AI model or its provider’s infrastructure was compromised.
Spain’s data-protection authority has received its first notification of a personal-data breach reportedly carried out with an autonomous AI agent powered by a large language model. The reported sequence is a warning that familiar security gaps can be exploited at machine speed.
The Agencia Española de Protección de Datos, or AEPD, says its information came from the affected organization’s notification and still needs further analysis. The investigation is ongoing, and little is currently known about the incident.
According to the AEPD account, the agent used publicly accessible files to log into a Spanish company’s system. It then reportedly scanned for vulnerabilities, exploited one, modified personal data and accessed invoices.
AEPD characterized the case as significant from a data-protection perspective because the agent reportedly chained several attack stages. The agency says AI can increase the speed, scale and adaptability of established malicious techniques, reducing the time defenders have to detect and contain an attack.
AEPD urged businesses to explicitly include AI-assisted and AI-driven attacks when assessing risks to personal-data processing. Francisco Pérez Bes warned that procedures designed for manually executed attacks may fall short when an agent can examine several assets, test different attack routes and rapidly adjust to what it finds.
The agency also highlighted digital identities and credentials. An agent using an account or API key can move across services at machine speed before anomalous activity is detected, it said. Spain’s National Cryptologic Center has recommended stronger baseline controls, faster vulnerability management, tighter identity protection, supplier oversight and governance for agent use.
Loading discussion...
Join the conversation
Explain what should change first, if anything.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.