Check Point Research identified a campaign called Yoba that distributes malware through counterfeit ChatGPT and Claude desktop apps. The fake projects—named “GPT-5.6 Sol,” “GPT-6 Astra” and “Claude Opus”—use authentic GitHub infrastructure and SEO-poisoned pages to rank among top Google results. Downloaded ZIP files contain a legitimately signed Setup.exe from OpenAI or Anthropic alongside a malicious winmm.dll. When run, the signed executable sideloads the DLL, deploying a dropper that pauses for about 12 days, skips systems with CIS-region keyboard layouts, and demands administrator rights. The dropper copies a Windows Media Player binary to ProgramData, creates a scheduled task authored by Microsoft Corporation, and then loads a second DLL that contacts a Telegram-based channel, queries Google DNS-over-HTTPS for encrypted instructions, and fetches an XMRig cryptominer from Catbox. Researchers warn future waves could deliver other malware families, and the use of mainstream services makes detection harder. 1
Published by CSO Online · On GETSECURE.ai ·
Fake ChatGPT, Claude apps spread malware via signed AI binaries
⚡ High Priority
Why This Matters
Developers and IT admins should verify the source of AI-related installers and monitor for unexpected DLL loading, scheduled tasks, and traffic to legitimate services used for command-and-control.
References
- CSO Online. (2026, October 28). Fake ChatGPT, Claude apps spread malware via signed AI binaries. CSO Online. https://www.csoonline.com/article/4241634/fake-chatgpt-claude-apps-spread-malware-via-signed-ai-binaries.html
Original Source
CSO Online
Read original →