What happened
The FBI issued a flash cybersecurity advisory warning that hackers are using autonomous AI tools to break into companies across the United States and around the world. The warning names Artex, an open-source AI penetration-testing platform, as one of the tools behind a string of recent attacks.
Artex connects to large language models from Anthropic and OpenAI, as well as open-source and non-U.S. models, and uses them to probe a target's network on its own. It scans for weak spots, then exploits simple flaws — weak protections and exposed or stolen passwords — to get a foothold. From there it spreads through the network hunting for databases, mailboxes and sensitive files.
The FBI says some Artex attacks move from first scan to full compromise "in under three hours." The tool is noisy: it throws off reams of traffic that can trip a company's defenses. But it moves so fast that human teams often cannot keep up.
The advisory lists financial services, government, tech, defense, phone and internet providers, and law firms among the industries hit in recent months. Two campaigns stand out: a widespread breach of customers of American IT firm ConnectWise, and attacks across South Korea's financial sector.
Why it matters
The ConnectWise breach shows how far one tool can reach. The FBI confirmed Microsoft's earlier finding that hackers used Artex to break into ConnectWise and reach the Microsoft 365 cloud environments of at least 2,250 of its customers. Microsoft says the hackers are likely Chinese, and that they stole data from affected customers. ConnectWise did not respond to a request for comment.
In South Korea, Artex attacks hit financial institutions that the FBI says had weak defenses and lacked protections such as multi-factor authentication. The personal data of at least 68,000 people was stolen.
Security firm Barracuda was also breached in an Artex-linked attack by financially motivated criminals. The tool found an exposed credential belonging to a third-party service that Barracuda had not revoked when it ended work with the vendor. That opened the door to stores of credentials Barracuda used to collect data from its customers' systems. Barracuda said its logs were not detailed enough to rule out access, so it assumed the worst and changed all of its customer credentials.
"Defenders who rely on spotting malicious activity and then responding to it will often arrive after the attacker has finished," wrote Barracuda chief information security officer Arve Kjoelen.
The speed is the story. Traditional hacking took days of manual work. Artex compresses the whole chain — recon, exploit, spread — into hours, and it runs on models anyone can rent. That puts small and mid-size companies, which often lack round-the-clock security teams, squarely in the crosshairs.
For American readers, the exposure is direct: the targets are U.S. companies, U.S. government offices and U.S. law firms, and the stolen data belongs to their customers and clients. Washington is already tightening controls on advanced AI technology — [Congress moved to restrict AI chip exports after the Nvidia scandal](/article/congress-tighten-ai-chip-export-rules-china-nvidia) — while AI systems spread deeper into critical infrastructure, including [the FAA's nationwide AI air traffic system](/article/faa-rolls-out-ai-air-traffic-system-nationwide). And the race to build autonomous AI agents keeps accelerating, with [LangChain hitting a $3 billion valuation](/article/langchain-hits-3b-valuation-ai-agent-race).
What happens next
Artex's developer has since made the software closed-source to stop further attacks, the FBI says. But the earlier open-source version keeps circulating and remains "widely accessible."
Worse, the tools themselves are leaking. Last week researchers found hundreds of Artex systems sitting exposed on the public internet with no password at all — likely misconfigured by their operators — including the very Artex system used to target South Korea.
The gaps the FBI cited in every case are basic ones: exposed credentials, missing multi-factor authentication, and vendor access that was never revoked. Companies that still rely on spotting an attack and then reacting will keep arriving after the damage is done.



