Spanish Rail Operator Renfe Confirms Cyberattack Involving AI, Exposing User Data Spanish state-owned railway operator Renfe disclosed a cyberattack that compromised user data, marking what local media reports as Spain’s first known AI-assisted breach targeting a public company. The incident, confirmed on Friday, originated from previously compromised servers belonging to railway infrastructure manager Adif, which were interconnected with Renfe’s systems. The attack exposed limited user information, primarily names and email addresses, though Renfe stated there was no evidence the data had been publicly leaked. The company also confirmed that no financial details, payment methods, or sensitive identification documents were accessed. Despite the breach, rail services remained unaffected, following weeks of attempted attacks that had been successfully blocked. Spanish daily El Mundo cited investigative sources claiming a criminal group used an AI system resembling Anthropic’s technology to target Adif’s website, which was temporarily unavailable. Reports indicated the breach lasted several days and resulted in the theft of approximately 500GB of data. While Renfe did not disclose the number of affected users or the exact timeline, La Razón suggested the attack bore the hallmarks of a foreign cybercriminal group. The incident adds to growing global concerns over AI’s role in cyber threats, following a series of hacking incidents involving models from OpenAI and Anthropic. Security experts have warned of potential hybrid attacks, including cyber operations, amid heightened geopolitical tensions, particularly in relation to Russia’s ongoing invasion of Ukraine.

Weekly Cybersecurity Roundup: AI as Target and Tool, Trusted Software Exploited in Attacks This week’s cybersecurity incidents highlight two key trends: artificial intelligence (AI) as both a target and an attack vector, and threat actors abusing trusted software to evade detection. ### Key Incidents Spain: Renfe Customer Data Breach via Compromised Partner Servers On September 25, Spain’s national rail operator Renfe confirmed a data breach exposing customer names and email addresses. The attack originated from compromised servers belonging to Adif, the state-owned rail infrastructure manager. While no financial data or train services were affected, the incident underscores risks from third-party supply chain vulnerabilities. Australia: Stolen AI Credentials Cost Firms Up to $600,000 The Australian Signals Directorate (ASD) warned that attackers are hijacking corporate AI services by stealing API keys, session tokens, and supplier access. Once inside, they drain paid credits, generate malicious content, and use AI agents to infiltrate connected systems. The financial impact has reached $600,000 per incident in some cases. Poland: Patient Data Theft via Medical Software Vulnerability Between August 22–23, attackers exploited an SQL injection flaw in Medyc, a Polish medical records platform developed by Qbusoft. The breach exposed patient names, national ID numbers (PESEL), addresses, and contact details dating back to July 2024. The intrusion went undetected for over two weeks, only discovered on September 8–9. Global: Phishing Campaigns Abuse Legitimate Remote Management Tools Microsoft tracked phishing campaigns tricking victims into installing MSP360, a legitimate remote management tool, disguised as Zoom downloads, Adobe updates, or delivery notices. After gaining access, attackers deployed ConnectWise ScreenConnect as a secondary persistence mechanism, blending malicious activity with routine IT operations. Europe: Police Chiefs Warn of AI’s Role in Expanding Cybercrime At Europol’s headquarters (September 28–30), over 500 police leaders from 59 countries discussed how AI is accelerating criminal operations. Threat actors are leveraging generative AI, deepfakes, and autonomous agents to scale fraud, cybercrime, migrant smuggling, and child exploitation. While AI also aids law enforcement investigations, its misuse is increasing the speed and scale of attacks. ### Emerging Threat Patterns - AI as a Target & Weapon: Attackers are hijacking AI services for financial gain and lateral movement, while criminals use AI to enhance fraud and cybercrime. - Abuse of Trusted Software: Legitimate tools from medical platforms to remote management software are being exploited to bypass security controls and maintain persistence. - Supply Chain Risks: Breaches like Renfe’s demonstrate how vulnerabilities in third-party infrastructure can expose customer data. These incidents reinforce the need for heightened scrutiny of AI credentials, remote access tools, and vendor security practices in enterprise environments.