OpenAI busts dozens of accounts in North Korea-linked remote hiring scam

OpenAI has banned dozens of accounts that were systematically abusing ChatGPT to run a deceptive employment scheme, and the playbook looks straight out of Pyongyang’s IT-worker fraud manual. The company’s February 2025 threat report details how these actors didn’t just use AI for one step—they weaponized it across the entire hiring lifecycle.

Starting with the application, operators generated tailored resumes, online profiles, and cover letters designed to match specific job listings perfectly. They then created entirely separate “support” personas to serve as fake references and internal referrals, building a web of artificial credibility. But the scheme didn’t stop at getting past HR filters. The report notes the applicants used ChatGPT to generate plausible real-time responses to both technical and behavioral interview questions, effectively using the model as a brain implant during the vetting process.

What makes this operation particularly sophisticated is the human infrastructure it recruited alongside the AI tools. The actors ran social media campaigns to find real people willing to receive company laptops at their home addresses or lend their identities to pass background checks. Once hired, the fraudulent workers kept the grift going—using ChatGPT to write code, handle IT troubleshooting, message colleagues, and craft elaborate cover stories for why they couldn’t turn on a webcam or why their VPN kept pinging a server in Southeast Asia.

OpenAI says its visibility into the full distribution of this content is limited, but confirmed some of the material was posted to LinkedIn. The activity mirrors the tactics, techniques, and procedures that Microsoft and Google have separately attributed to a North Korean state-directed effort to funnel hard currency into the regime by placing remote IT workers in Western companies. While OpenAI can’t confirm the actors’ locations or nationalities, the overlap in tradecraft is striking. The company has shared its findings with industry peers and relevant authorities, and the banned accounts represent a disruption of a network that almost certainly extends far beyond what any single platform can see. This case is a stark reminder that generative AI’s utility for fraud isn’t theoretical—it’s operational, and it’s scaling the classic resume scam into something far harder to spot.