Anthropic explains how North Korea is misusing AI to infiltrate IT companies

North Korea is systematically misusing artificial intelligence (AI) for the benefit of its regime, particularly in cybercrime and illicit foreign currency acquisition activities. A report published by Anthropic revealed that North Korean IT engineers are using an AI model called 'Claude' to infiltrate major technology companies as remote workers with false resumes and maintain fraudulent employment.

Detecting and countering misuse of AI: August 2025 \ Anthropic
https://www.anthropic.com/news/detecting-countering-misuse-aug-2025

North Korean operatives are using AI in a wide range of tasks, overcoming technical and language barriers that previously hindered their activities. For example, they are using AI to create believable resumes, self-promotional materials, and even technology portfolios to deceive recruiters.

FBI warns that North Korean IT workers employed under false identities are stealing confidential information and blackmailing their employers - GIGAZINE

When candidates undergo technical assessments during hiring, AI is used in real time to generate answers for coding tests and technical Q&A sessions, disguising advanced technical skills that they may not actually possess. Even after being hired, AI is relied upon in every aspect of the job, including daily development tasks, team communication, and responding to code reviews. This allows operatives who would not normally have basic coding skills or specialized English conversation skills to retain technical positions.

There have been reports of North Korean workers actually working remotely for American companies.

An Arizona woman was sentenced to eight and a half years in prison for running a 'laptop farm' that helped North Korean workers earn over ¥2.5 billion by hiding their identities and working remotely at 309 companies.

In the past, North Korean IT operatives had to undergo years of specialized training at specialized institutions within the country, and the lack of training capacity was a bottleneck for their activities. However, with the advent of AI, even personnel without advanced training can now be deployed as immediate assets, dramatically expanding the scale of their activities.

The FBI estimates that the revenue earned from these activities amounts to tens of billions of yen per year, and that this money is believed to be allocated to North Korea's weapons development program. The use of AI not only supplements the capabilities of individual operatives, but also eliminates bottlenecks in training, making it possible to send more operatives to companies around the world and dramatically expanding the scale of the entire operation.

This new method, which could be described as 'simulating capabilities using AI,' marks a new stage in state-sponsored illicit activities in cyberspace and poses a serious threat to the international community. While Anthropic has taken measures such as suspending accounts found to be engaging in such misuse and strengthening its monitoring system, there are concerns that AI misuse will become even more sophisticated and large-scale in the future.