Canada’s cybersecurity agency says agentic AI is accelerating cyberthreats on an industrial scale, after a Chinese state-aligned hacker used the technology to automate attacks against hundreds of organizations around the world.
The Canadian Centre for Cyber Security said in an advisory Monday that recent activity by a threat actor called Storm-0940 offers an early look at how artificial intelligence is “reshaping cyber operations.”
The activity made use of agentic AI – an AI system that can make decisions and use external tools with limited human supervision or instruction.
Microsoft said in a recent blog post that since as early as September, Storm-0940 has been using an agentic penetration testing system called ARTEX to automate attacks, some of which have resulted in an organization’s cloud infrastructure being compromised. In some cases it took less than three hours, Microsoft said. (Penetration testing, or pen testing, is a security exercise in which a cybersecurity expert attempts to find and exploit vulnerabilities in a computer system for defensive purposes.)
“Agents are not constrained by factors affecting humans – they can work around the clock, on weekends, and on holidays and attempt multiple breach paths simultaneously, enabling significantly reduced time from selecting a target to achieving objectives,” Microsoft wrote. Organizations across a broad range of industries were targeted, including financial services, manufacturing, defence, automotive, media and technology.
Bell, Cohere release AI model designed to aid cybersecurity investigations
The warnings are the latest development in the escalating AI arms race between attackers and defenders, both looking to take advantage of the technology’s capabilities.
Last week, security researchers using AI technology disclosed the discovery of critical vulnerabilities in PHP, an open-source programming language used to run countless websites globally. Many global government websites were temporarily taken offline as developers raced to patch the flaws.
The Canadian Centre for Cyber Security, which is part of the Communications Security Establishment Canada, issues alerts and advisories relating to cyberthreats, vulnerabilities or incidents.
The centre said in its latest advisory that AI has already made it easier for hackers to find weaknesses, develop malicious code and attack many organizations simultaneously. Agentic AI accelerates this trend by linking multiple stages of an attack together, the centre said.
“As automated agents become more capable, organizations may face attacks that move across several systems at once,” the centre cautioned. “Machines can act faster than many existing processes and policies were designed to handle. Decisions that once allowed time for discussion may soon have to be made within minutes.”
It also noted that although AI-enabled cyberattacks are detectable by sophisticated defenders, as the technology improves its malicious activity will likely become more targeted and harder to spot.
“We do see AI enhancing and being used across different elements in the kill chain, which is the steps that threat actors take,” Rajiv Gupta, the agency’s head, said in an interview last week at the Cisco Connect Canada 2026 event in Toronto.
Canada’s plan to counter the U.S. and China in AI is taking shape. Will it work?
Cybercriminals are using AI to create deepfakes – AI-generated images, audio or video – or more convincing e-mails for victims to click on, automate the exploitation of vulnerabilities in code and analyze exfiltrated data, said Mr. Gupta, calling recent developments “a sign of things to come.”
The agency noted that agentic AI does not render established cybersecurity practices obsolete, but rather makes them more critical.
“Weakly protected accounts, excessive access, exposed systems, unpatched software and gaps in monitoring remain common ways into an organization,” the centre said.
It urged organizations to take steps to address weaknesses, such as patching online systems promptly and reviewing the access granted to third parties and cloud applications.
“AI increases the consequences of weak controls and slow decisions, but the fundamentals of good cybersecurity remain effective,” the advisory said.