Washington, D.C.: A sophisticated phishing campaign powered by artificial intelligence has compromised hundreds of organizations worldwide, researchers at Huntress have revealed. The campaign exploited Microsoft cloud accounts using unique phishing lures, which experts suspect were generated with AI tools.
Rich Mozeleski, product manager for Huntress’ identity team, told CyberScoop that while the campaign currently traces back to a relatively small actor and roughly a dozen IP addresses, it has managed to compromise hundreds of targets over recent weeks. The attack pace accelerated sharply from March 3, after initially affecting a few dozen victims daily.
FCRF Launches Premier CISO Certification Amid Rising Demand for Cybersecurity Leadership
“Just the volume of attacks was staggering, and the effectiveness was unprecedented,” Mozeleski said. Unlike typical phishing campaigns, the emails did not repeat domains or content, indicating the use of AI to generate unique lures at scale. Techniques ranged from conventional email prompts to QR codes and co-opted file-sharing sites.
The campaign specifically targeted Microsoft’s authentication flow on devices such as smart TVs, printers, and terminals. This allowed attackers to obtain valid OAuth tokens for up to 90 days without needing passwords or multi-factor authentication. Huntress said it was able to prevent post-compromise activity in its customer base, but the total number of victims could reach into the thousands, far exceeding the 344 detailed in their report.
Affected sectors included construction, trade, law firms, nonprofits, real estate, manufacturing, finance, insurance, healthcare, and government/public safety organizations. To mitigate the threat, Huntress issued a conditional access policy update to 60,000 Microsoft cloud tenants, blocking emails from Railway-associated domains — an unprecedented step according to Mozeleski.
Railway’s platform exploited for phishing infrastructure
Researchers noted that attackers leveraged Railway’s Platform as a Service (PaaS), originally designed to help non-coders deploy websites and tools, to spin up phishing infrastructure. By using compromised domains and AI-generated lures, the campaign bypassed most commercial email filters. All observed attacks originated from Railway.com IP infrastructure.
Railway solutions engineer Angelo Saraceno confirmed the company was notified on March 6 and had banned associated accounts and blocked the domains. He acknowledged the challenge of balancing fraud detection with minimizing false positives and cited a February incident where overzealous automated enforcement caused customer outages.
Mozeleski suggested that Railway, like other SaaS trial providers such as MailChimp and HubSpot, could implement stricter vetting and resource usage controls to prevent misuse for cyberattacks.
AI levels the playing field for smaller cybercriminals
Security experts see the campaign as a warning that low-level cybercriminals are becoming significant beneficiaries of generative AI. Prakash Ramamurthy, Huntress’ chief product officer, said, “We are seeing crooks as the first movers of AI. They exploit personally identifiable information and model training with no qualms, and this campaign shows the speed at which AI amplifies their attacks.”
The incident highlights the evolving threat landscape, where AI-generated infrastructure and content allow smaller groups to emulate tactics previously seen only in advanced, state-sponsored campaigns. Experts warn that organizations need to adapt cybersecurity policies and monitoring to counter AI-driven phishing threats.
Huntress’ findings underline the urgency of conditional access policies, multi-factor authentication, and AI-aware threat intelligence, particularly in cloud environments, to mitigate the increasing risk posed by automated, scalable phishing attacks.
