Microsoft reported that during the ConnectWise Cloud Backup compromise revealed last week that affected Microsoft environments, Chinese threat actor Storm-0940 used AI penetration-testing tool Artex to accelerate scanning and reconnaissance, password spraying and OAuth application abuse.
According to an Oct. 21 post from Microsoft, Storm-0940 used Artex to target hundreds of Microsoft environments across multiple industries, including aerospace and defense, automotive, chemicals, energy, financial services, government, legal, manufacturing, media, the military, technology, and think tanks and non-profits.
"Artex is another example of AI-powered offensive security tools being adopted by attackers,” said Alex Wells, head of product strategy at Hadrian. “Earlier this year, we saw CyberStrikeAI associated with a campaign targeting hundreds of Fortinet firewalls. Now we’re seeing Artex used by a state-aligned threat actor to automate reconnaissance, credential attacks and the abuse of trusted cloud applications.
Wells said what makes Artex a concern was its ability to coordinate multiple AI agents across different stages of an attack, with relatively little human intervention. In the ConnectWise campaign, Wells said Microsoft observed activity reaching more than 2,250 cloud tenants, with more than 2,900 sign-in attempts per minute. In several attacks, the time from initial targeting to privileged access was less than three hours.
“These tools are beginning to change the economics of cyberattacks,” said Wells. “Activities that previously required considerable time and specialist expertise can increasingly be automated and executed in parallel. That lets attackers target more organizations, test more attack paths, and exploit opportunities much faster.”
Shane Barney, chief information security officer at Keeper Security, said the integration of agentic AI frameworks into nation-state cyber operations represents a fundamental shift in the speed and scale of supply chain attacks.
Barney said when a threat actor can use an autonomous AI agent to automate reconnaissance, execute password sprays and abuse OAuth applications to gain privileged cloud access in under three hours, the traditional window for detection and response effectively disappears. This is a critical vulnerability given that research from 2026 reveals more than 72% of organizations take hours, days or even weeks to detect credential misuse, Barney continued.
“This campaign highlights a structural vulnerability in how modern environments operate,” said Barney. “The attackers weaponized the implicit trust granted to downstream applications and used AI to rapidly navigate the identity layer. Autonomous agents thrive in environments where non-human identities are loosely governed and standing privileges are left unchecked. When service principals and OAuth apps hold persistent, high-level access without continuous monitoring, they provide a frictionless path for AI-driven lateral movement.”
Anthony Pillitiere, co-founder at Horizon3, added that while the ConnectWise breach has been framed as an AI story, AI didn't break in — a poorly protected vendor secret with access to thousands of tenants did, and agentic tooling just made the damage happen faster.
Pillitiere said preventing the next one takes third-party risk management that makes vendors prove their security instead of describing it, starting with how they protect the keys to our environments.
“And for every organization downstream, the lesson is to assume breach,” said Pillitiere. “Treat each privileged third-party app in your tenant as a path an attacker may already control, watch for credential and role changes, and [continuously] test what that access could reach before someone else does.”
