North Korea’s devious use of technology in its cybercrimes dominate global headlines: from stealing $6.3 billion in cryptocurrency from a digital wallet in a mere 20 minutes, to the U.S. Treasury Department sanctioning 11 individuals (some of them Russians) for defrauding U.S. companies of $800 million in salary payments. As the Treasury stated this week:
Today’s action targets key facilitators and financial networks that help sustain these schemes and move illicit proceeds on behalf of the DPRK. These networks have leveraged sophisticated cryptocurrency transactions, money laundering techniques, and emerging technologies, including artificial intelligence tools, to generate and transfer millions of dollars for the DPRK government. In addition to generating revenue, DPRK IT workers have exploited their access to corporate networks to deploy malware, exfiltrate sensitive data, and steal proprietary information, creating significant risks for businesses worldwide.
As we know, it’s all to prop up the DPRK and its nuclear missile program. While producing my documentary, “The Only Human in the Room,” I can attest firsthand to their growing use of sophisticated AI tools to brazenly conduct these schemes.
The U.S. Treasury points directly at Russian intermediaries leading IT worker fraud on behalf of the DPRK. This could be why one of the North Korean operatives we encountered in a live Zoom job interview had used AI to face swap with a Russian dating profile. I was able to trace that facial pattern using open-source tools and the Russian search engine Yandex.
In a recent film shoot last month with Pindrop investigator Katelyn Halbert who interviewed over 15 job candidates in one day, she concluded that this HR marathon led to only one genuine candidate. The rest were fraudulent identities, and almost all foreign state agent actors. Exasperated, Katelyn finally challenged three suspicious interviewees to say something negative about Kim Jong Un, two of them hung up immediately.
One got clever and instead proclaimed “I don’t like pineapple pizza” instead of expressing dislike for the North Korean dictator.
Microsoft Threat Intelligence’s latest report on North Korean threat actor tradecraft notes an expansion of use of AI “as a force multiplier across their operations, particularly to accelerate and scale established workflows” but there’s little evidence of the use of agentic AI to “enable autonomous operations inc.” That’s the next frontier for where I’m headed with the film, but in the meantime, note the preference of North Korean actors for synthetic media generation, “as it significantly enhances existing social engineering capabilities.”
AI might be their new weapon of choice, but the ultimate vulnerability remains human nature. Because while a deepfake can perfectly map a stolen face in real time, it still hasn’t figured out how to casually take an insult about the Supreme Leader.


