Microsoft: Hackers used open-source tool ARTEX for AI attacks on hundreds of organizations

What this is about

A Japanese news article (Yahoo Japan, via Bing News, October 23, 2026) reports that hackers used AI to attack several hundred organizations worldwide. It cites Microsoft. The source is only available as a short excerpt, so many details are unknown: who the attackers were, which organizations were affected and how successful the attacks were. The source says nothing on these points.

The tool: ARTEX

The excerpt names ARTEX as the tool used. It is an open-source tool built for penetration testing. These are authorized checks in which security professionals probe a company's networks and systems for weaknesses before real attackers do.

ARTEX deploys multiple AI agents based on large language models (LLMs). An AI agent is a program that plans and carries out multi-step tasks on its own. According to the report, the agents automatically handle:

• gathering information about the target,
• searching for vulnerabilities,
• executing attack methods.

How autonomous is it?

According to Microsoft, ARTEX needs almost no human operation. Depending on configuration, the tool can carry out the work all the way through to data exfiltration. In other words, a defensive tool can be turned into an attack tool with little effort.

Context

The case illustrates the familiar dual-use problem: open-source security tools can help defenders and attackers alike. With AI agents, attacks get cheaper, since they need less expertise and manual work. Still unclear: how many attacks succeeded, who is behind them and what exactly Microsoft published. The source is a short second-hand excerpt, and Microsoft's original report was not available. Treat the claims with caution.