Fake ChatGPT and Claude Apps on GitHub Are Spreading Yoba Malware

Check Point Research has spent two months tracking a campaign that exploits the popularity of AI tools such as ChatGPT and Claude. Attackers created fake GitHub projects posing as AI desktop applications, including GPT-5.6 Sol, GPT-6 Astra and Claude Opus. Because they sit on real GitHub infrastructure, the projects look convincing to potential victims.

The researchers say the repositories do not show fake stars or forks, yet they still appear among the top Google results. They believe this is most likely SEO poisoning, where attackers tune titles, descriptions and hosting domains to rank highly in search engines. GitHub.io pages describe the supposed software, and download links point straight to malicious ZIP files hosted as GitHub releases. Each repository used a different email address and a different apparent timezone, which likely reflects a deliberate effort to hide where the attacker really operates. The first repository was created around July 2026, with malicious releases uploaded at the end of August and a third about five days later.

Inside the installers sits a malicious component called the YobaDropper, which contains another component, the YobaLoader. The loader contacts attacker-controlled infrastructure and fetches the final payload, which is used to run unwanted software on infected systems. This summary draws on the opening section of the report only, so the full research may contain further detail.