New Wave of DPRK Attacks Uses AI-Inserted npm Malware, Fake Firms, and RATs

New Wave of DPRK Attacks Uses AI-Inserted npm Malware, Fake Firms, and RATs
Researchers discovered malicious code embedded in the npm package β€œ@validate-sdk/v2” that acts as a credential and crypto-wallet stealer when introduced as a dependency to other projects. ReversingLabs calls the campaign PromptMink and links it to North Korean actor Famous Chollima, which leverages AI-generated code, layered dependencies, typosquatting and transitive-dependency tricks to deliver RATs and exfiltrate secrets. #PromptMink #FamousChollima

Keypoints

  • A malicious npm package β€œ@validate-sdk/v2” was used to plunder sensitive secrets and crypto-wallet credentials.
  • ReversingLabs attributes the PromptMink campaign to North Korean actor Famous Chollima (Shifty Corsair), linking it to Contagious Interview/Trader activity.
  • Attackers used AI-generated code (Anthropic’s Claude Opus) and a two-layer dependency strategy where benign first-layer packages pull malicious second-layer packages.
  • The campaign abuses transitive dependencies, typosquatting, GitHub release artifacts and Vercel-hosted C2 to evade detection and persist.
  • Related operations (OtterCookie, Contagious Trader, graphalgo) deploy RATs, steal AWS/GitHub keys and lure developers via fake companies and job interview scenarios.

Read More: https://thehackernews.com/2026/04/new-wave-of-dprk-attacks-uses-ai.html