Yoba Campaign Abuses Legitimate AI Apps to Deploy Crypto Miners
Original source
Fake ChatGPT and Claude Apps Abuse Signed Executables to Deploy Crypto Miners
The Hacker News →Threat actors are distributing cryptocurrency mining malware through fake desktop applications for ChatGPT and Claude. The campaign, called Yoba, uses counterfeit GitHub repositories claiming to be tools like GPT-5.6 Sol, GPT-6 Astra, and Claude Opus. Each repository contains ZIP files with a legitimate signed executable from OpenAI or Anthropic alongside a malicious DLL file, leveraging DLL sideloading to bypass security controls. The attackers created companion GitHub.io pages and used SEO poisoning to rank malicious repository links high in Google search results.
The malware incorporates sophisticated evasion techniques to avoid detection and analysis. It performs 200,000 iterations of delays totaling 12.73 days and skips infection on systems with keyboard layouts from Russia, Belarus, Tajikistan, Armenia, Azerbaijan, Georgia, Kazakhstan, Kyrgyzstan, Turkmenistan, and Uzbekistan, along with systems using the Tatar language. The loader component generates domains via a domain generation algorithm seeded from images downloaded from a Telegram channel, creates domains in the format “adjective-noun-verb-digit.dedyn[.]io,” and queries Google’s DNS-over-HTTPS API to retrieve command data. The final payload injects into svchost.exe to execute an XMRig cryptocurrency miner.
The first repository appeared around July 2026, though malicious versions weren’t uploaded until August. The campaign exemplifies how attackers exploit public interest in AI applications and legitimate platforms like GitHub to distribute malware that appears trustworthy initially.
Read the full article
Continue reading at The Hacker News →This is an AI-generated summary. Read the original for the full story.