European Commission Confirms an AI-Assisted Attack on Its Systems
· Cortex Solutions

The European Commission confirmed that AI tools were used in an attack on its systems. Here is what is known, what is not, and the lessons for Saudi firms.
The European Commission confirmed on Thursday, November 19, 2026 that AI tools were used in a cyberattack on a number of its internal systems. The Brussels Times that day quoted Commission spokesperson Thomas Regnier: "Indeed we can confirm that this was done with AI tools." The confirmation came after Politico broke the story that morning.
This article relies on three reports published on November 19: The Register, The Brussels Times and Techzine. Some of their details come from Politico's reporting.
What happened?
According to The Brussels Times, the spokesperson said the attack was detected on October 22, and the disruption mainly hit internal tools, especially those used for human resources. All three sources agree on two affected systems:
- Sysper, which staff use for payslips, timesheets and leave requests.
- MyPMO, which handles medical reimbursements.
The Register reported that both systems were taken offline and the disruption lasted more than ten days. They came back earlier in the week, before Sysper went down again on Wednesday. The Register and Techzine add that translation services were also affected. On salaries, The Register, citing Politico, said they were paid on time, while Techzine said only that most were paid.
What is ARTEX?
An EU official told Politico that the cybersecurity emergency response team CERT-EU identified the use of a tool called ARTEX. The Register describes it as a publicly available AI-powered penetration testing framework. The Brussels Times and Techzine describe it as a publicly available tool for finding vulnerabilities in IT systems.
According to The Register:
- CERT-EU warned on November 5 of attacks it believed were ARTEX-assisted.
- The tool's developer pulled its GitHub repository in October, but forks remain available.
- The FBI pointed to more than 1,600 ARTEX consoles exposed to the internet.
The Register also reported the view of Snehal Antani, CEO of Horizon3.ai, that AI frameworks let a single actor run fleets of agents with little human input and make attribution harder.
What is still unresolved?
It is important to separate what is confirmed from what is possible:
- Data: the Commission says, per The Brussels Times, that it found no evidence of a breach affecting its staff as a whole. The Register and Techzine, by contrast, speak of possible exposure of personal data and note that it is not yet known whether staff data was compromised.
- The tool's role: Techzine says it is unconfirmed whether ARTEX enabled the breach itself or was used only for reconnaissance.
- The attacker: the Commission has named no one, and the investigation continues.
- The scale of damage: Techzine quoted an unnamed EU official as saying the Commission believes it has contained the bulk of the attack, but the full extent of the damage is still unclear.
The spokesperson declined to detail countermeasures because disclosure could undermine their effectiveness, according to The Brussels Times. The Commission said it had taken the necessary steps and informed those concerned.
A lesson in internal communication
Techzine reported a point that matters as much as the technical one: three civil servants complained about poor communication, because an IT alert about Sysper referred to a general problem without mentioning a cyberattack. Long silence does not stop an attack, but it can leave employees without clear instructions when they need them.
What does this mean for Saudi businesses?
The sources we read mention no affected Saudi or Gulf organization. What follows is Cortex's reading and recommendations, not information from the investigation.
First, the systems that went down were not secret or exotic. They handle payroll, leave and reimbursements, and almost every company has them. Second, automated vulnerability-finding tools are now widely available, so a small company should not assume its size protects it.
Practical steps we recommend:
- List your internet-facing systems and close what you do not need.
- Speed up security updates, starting with HR and finance systems.
- Turn on two-step verification and give each employee only the permissions their job needs.
- Keep isolated backups and actually test restoring them, not just on paper.
- Prepare a fallback for paying salaries and recording attendance if a system is down for ten days.
- Write an incident message for staff and customers in advance, and decide who approves it.
- Review with your adviser your legal obligations on personal data protection and incident reporting.
If you are building custom systems or bringing AI into your operations, make permissions and audit logs part of the design from day one. Learn about Cortex custom software: https://cortex.sa/services/custom-software
Sources
- The Register, Carly Page, November 19, 2026.
- The Brussels Times, Camille Guerry, November 19, 2026.
- Techzine, Colin Baak, November 19, 2026.