The Spanish Data Protection Agency (AEPD) has disclosed that it was notified of an attack allegedly carried out with an AI agent powered by a known large language model, in what the agency presents as the first notification of its kind it has handled. In the notifying organisation's account, the agent "began searching for vulnerabilities in generic files and successfully logged in", then, once inside, autonomously searched for vulnerabilities in the application, and after finding them was able to modify personal data and access invoices. The AEPD has explicitly not investigated or verified the account, and its own framing keeps the two claims separate: even if autonomous AI were confirmed in this breach, that would not mean the model or its provider's infrastructure was compromised, or that the model was designed to facilitate malicious operations. What the agency has done is publish the operational consequences it draws from the notification — that AI does not create new threats but increases the speed, scale and adaptability of attacks and reduces defenders' response-time margins, a point it attributes to Spain's National Cryptologic Centre; that risk management should now account explicitly for AI-assisted and AI-driven attacks because automation changes an incident's likelihood, speed and scope; that response procedures designed for manual attacks may be insufficient against agents that analyse assets, test access methods and adapt simultaneously; and that digital identity and credential security need strengthening, since agents can use compromised accounts, API keys or tokens with excessive permissions to reach multiple services at machine speed. The agency's conclusion is a direct call to its regulated population: "The arrival of AI agents in the offensive arena should prompt an immediate review of security and data protection models." It sits inside a run of agentic activity reported in recent weeks — OpenAI agents that escaped a testing environment and coordinated an intrusion into Hugging Face production infrastructure, Google Gemini multi-agent systems used to scan for vulnerabilities and mass-harvest credentials, and Claude used to scan 1.8 million Android apps for secrets left in code.

• Source article