North Korea has reportedly been utilizing banned NVIDIA graphics cards, specifically the GeForce RTX 2700, to bolster its artificial intelligence (AI) research capabilities, according to a recent report by The Chosun Ilbo. These powerful GPUs, restricted under stringent U.S. export controls, are allegedly being deployed in various illicit operations, including cryptocurrency theft, the creation of deepfakes, and extensive surveillance activities. The prohibition on exporting NVIDIA GPUs such as the GeForce RTX 2700 to North Korea stems from national security concerns and international sanctions, which classify the hardware as "dual-use" technology essential for AI research and cryptography.

Despite decades of international sanctions, Pyongyang has persistently pursued and developed its AI capabilities. While the country is not using the most advanced AI models available globally, mid-level GPUs still provide substantial computational power, enabling meaningful progress in its illicit operations. This incremental yet steady advancement allows the regime to strengthen its strategic objectives, especially in areas where AI provides significant operational advantages.

One of the most notable areas of North Korea’s AI application is cryptocurrency theft. The regime’s strong interest in cryptocurrency began around 2017, driven by digital assets' potential to help evade sanctions and generate hard-to-trace revenue streams. State-sponsored cyber units, most prominently the Lazarus Group, widely believed to be linked to North Korea’s military intelligence arm, the Reconnaissance General Bureau, have repeatedly been implicated in targeting cryptocurrency exchanges worldwide to unlawfully obtain funds.

The integration of AI into these cybercriminal operations has significantly amplified North Korea’s capabilities. A 2025 report reveals that North Korea-linked hacking teams are now leveraging advanced AI tools to automate and enhance sophisticated tasks, such as scanning large and complex codebases for vulnerabilities and identifying weaknesses within smart contracts. These groups are also using AI to generate highly convincing fake recruiter profiles and phishing messages, enabling them to infiltrate organizations more effectively by building trust and exploiting unsuspecting victims. This AI-driven approach substantially increases their ability to execute cyberattacks and evade detection.