OpenAI has banned numerous accounts involved in a deceptive employment scheme, potentially linked to North Korea. The activity mirrors tactics used by IT worker schemes that aim to funnel income to the regime through fraudulent hiring at Western companies.

The actors utilized OpenAI's models to generate content for various stages of the recruitment process, including résumés, cover letters, and online profiles for fictitious job applicants. They also created "support" personas for references and referrals, and crafted social media posts to recruit real individuals to aid their schemes. These real individuals were asked to host laptops or lend their identities for background checks.

"Job applicant" personas reportedly used the models to generate responses for interviews and to perform job-related tasks like writing code and messaging colleagues after gaining employment. They also devised cover stories to explain unusual behaviors, such as avoiding video calls or accessing systems from unauthorized countries. OpenAI banned dozens of accounts and shared information with industry peers and authorities to combat these threats.