A campaign called LLMShare exploits ChatGPT's content-sharing feature to display fake OpenAI outage notices directing users to download malware disguised as the ChatGPT desktop application. The attack begins with Google ads that redirect users searching for ChatGPT to legitimate shared pages on chatgpt.com, where custom HTML renders an outage message claiming the web version is unavailable. The fake notice includes a download button leading to openew[.]app, a site impersonating OpenAI's desktop app portal that uses cloaking to display malicious content only to targeted victims while showing harmless content to security scanners.
The website offers both macOS and Windows downloads that install malware, with earlier campaigns abusing similar AI platform sharing features distributing infostealers. Push Security noted the fake pages include "Show code" and "Remix with ChatGPT" controls, revealing the content is generated from custom HTML rendered by a ChatGPT prompt. The researchers also observed attacks abusing Claude Artifacts to host ClickFix-style lures tricking users into executing malicious commands.
Similar attacks earlier this year used Google advertisements to direct Claude download seekers to shared conversations containing malicious installation instructions. Other campaigns abused shared ChatGPT and Grok conversations impersonating software installation guides that instructed victims to execute commands installing malware. The abuse of AI platform sharing features represents an evolving tactic for distributing malware through legitimate and trusted domains. BleepingComputer's Windows version test revealed the malware executes various commands to detect virtual machines or legitimate computers.
Read more...