Alleged AI-aided cyberattack exploits assistant interface to breach German web host

Hackers allegedly assisted by AI breached a German web hosting company through a tool designed to allow customers to connect AI assistants to their accounts, the company disclosed over the weekend.

Mittwald CM Service, a hosting provider for web agencies based in North Rhine-Westphalia, disclosed in a blog post Saturday that hackers had compromised 56 customer projects after going undetected on its systems for eight days.

The company’s managing director Robert Meyer said the attackers exploited a flaw in Mittwald’s Model Context Protocol (MCP) server used to allow customers to connect their AI assistants to mStudio, a platform for managing websites and hosting services. It is unclear whether the flaw was in the company’s own code or in the open-source MCP software libraries.

Meyer said the attacker appeared to have used AI tools in the breach, explaining the company’s logs “show that he worked his way through our system extremely quickly” and that “in some cases, only seconds passed between steps that would take significantly longer for a human attacker.”

Mittwald provided no other evidence of AI involvement. The company did not immediately respond when asked to distinguish between the detection of AI tools and conventional scripts that also allow hackers to move faster than an unaided human operator.

According to Mittwald’s own timeline, the attacker first tested the MCP flaw amid a global scramble in October to apply patches for PHP2Shell. Meyer said “the attacker disguised his access attempts and carried out various diversionary tactics.”

“External IT forensic experts are continuously supporting us with the analysis. We have informed the relevant authorities,” Meyer said.

After forensic evaluations of the 56 compromised projects, and repeated scans of Mittwald’s own systems, the company said that as of Friday it could not detect any additional attacker activity.

“First the PHP2Shell vulnerability, now the attack on our systems. AI attacks will become more frequent in the future,” Meyer wrote. “We want to be transparent about this, and therefore we will publish a detailed account of the events next week, as soon as our analysis and recovery are complete. We will be working on this throughout the weekend and beyond.”

Alexander Martin

is the UK Editor for Recorded Future News. He was previously a technology reporter for Sky News and a fellow at the European Cyber Conflict Research Initiative, now Virtual Routes. He can be reached securely using Signal on: AlexanderMartin.79