European Commission hit by AI-assisted cyberattack

Artex attack disrupted internal systems for more than 10 days

An AI-assisted cyberattack forced the European Commission to take several internal applications offline last month, according to Politico.

Although the Commission believes it stopped most of the attack, the disruption still lasted more than 10 days and the data exposed is unclear.

The attack is said to have affected HR system Sysper, used to manage salaries, timesheets and leave, as well as medical reimbursement and translation services. Most affected applications are now back up and running, although at least one health insurance service is still unavailable.

The EU's own cyber service, CERT-EU, detected the campaign on 22nd October and is said to be pointing the finger at Artex as the attack vector.

Artex is a publicly available open source AI-powered pentesting framework, designed to find vulnerabilities so defenders can fix them.

However, attackers have noticed that its ability to find vulnerabilities works quite well for them too.

Commission spokesperson Thomas Regnier said investigators have so far found no evidence of a data breach affecting staff "as a whole," although the investigation continues and the organisation has yet to attribute the attack. There was apparently no ransom demand.

From pentesting to breaking and entering

Artex uses AI agents to automate elements of vulnerability discovery and exploitation. The original project was pulled from GitHub in October following reports of malicious use, but copies are still available.

CERT-EU warned this month that Artex's availability could enable more frequent and scalable automated attacks. Indeed, it has already been used in several cyber campaigns.

The FBI subsequently said criminals and state-sponsored groups were using the framework, with some attacks moving from initial targeting to privileged access in less than three hours.

ISACA's Chris Dimitriadis said artificial intelligence means attackers can now "hack at the speed of intent", arguing that organisations need to deploy the technology defensively just as aggressively.

Amanda Brock, CEO of OpenUK, told Computing that the breach was not reflective of AI itself, but more likely down to "poor management and guardrails" at the Commission.

Amanda added, "It is also reflective of the fact that all software, all AI, whether open or closed, has security risks. That means a digital world has security risks we need to meet and manage."

AI has not made cyber defence impossible, but it has shrunk an already tight window between a vulnerability becoming exploitable and somebody deciding to exploit it.

And this is not the Commission's first security problem this year. It suffered separate incidents affecting mobile-device infrastructure in January and its cloud environment in March.

The Commission has not said exactly what data was targeted and if any breach occurred. HR systems often contain names, addresses, pay information and other personal details useful for follow-up attacks.