Fake ChatGPT and Claude desktop apps on GitHub deliver crypto miners via signed executables
The trust placed in signed OpenAI and Anthropic executables is what lets the Yoba campaign's malware slip past some security tools.
A campaign called Yoba is distributing a malware family, YobaLoader, through fake GitHub projects posing as desktop apps for ChatGPT and Claude. The repositories bundle a legitimate signed OpenAI or Anthropic executable alongside a malicious winmm.dll, so the malware loads through DLL sideloading and evades some security tools. Check Point Research reports the malicious loader ultimately injects an XMRig cryptocurrency miner into svchost.exe, with command-and-control reached via Google DNS-over-HTTPS and file-hosting services. Early versions excluded Russia-linked keyboard layouts and carried heavy anti-analysis checks that were later removed, and the first repository is believed to date from around July 2026 with malicious releases uploaded at the end of August. Developers and organisations should treat unofficial AI desktop installers from GitHub with suspicion, as the campaign relies on the trust placed in signed executables and well-known product names.
Why it matters
Anyone who downloads unofficial AI desktop installers from GitHub can end up running a cryptocurrency miner on their machine without noticing, because the malicious loader sits beside a genuine signed program and hides its traffic inside ordinary-looking DNS requests. The campaign shows that a known product name and a valid signature are not enough to establish that a download is safe, and that a lack of visible symptoms does not mean a system is clean.
Matters or hype?
Does this story matter, or is it hype? Vote before you see what everyone else said.