Malware · 5h ago
Check Point Research tracked a two-month campaign in which fake ChatGPT- and Claude-themed GitHub projects distributed the Yoba malware chain. The projects posed as desktop apps, including GPT-5.6 Sol, GPT-6 Astra, and Claude Opus, with download ZIPs hosted as GitHub releases.
Their GitHub Pages described the supposed software and linked to the ZIPs. Despite showing no signs of fake stars or forks, the repositories appeared near the top of Google results, likely through search-engine optimization (SEO) poisoning. Inside, YobaDropper carries YobaLoader, which contacts attacker-controlled servers to fetch the final payload. Familiar branding and search placement make a third-party download look credible; this is deceptive distribution, not a flaw in ChatGPT or Claude.
For organizations whose staff seek unofficial AI desktop clients, exposure starts at software discovery and can continue after the ZIP runs, when the loader fetches another payload. The report does not establish how many people installed the apps or what the final payload did, leaving the campaign’s reach and downstream impact unclear.
1 source covering this story
Tricks from the Past deployed via signed ChatGPT and Claude binaries - Check Point Research
Key Points Introduction Over the past two months, we have been tracking a campaign that abuses the popularity of AI tools such as ChatGPT and Claude to spread a custom malware chain we call Yoba.
Part of the PlainSec briefing for 2026-10-27