ChatGPT Share Links Abused to Host Fake Outage Pages That Deliver Malware
Threat actors are exploiting ChatGPT's share link feature to display fake OpenAI outage pages that trick users into downloading malware disguised as the ChatGPT desktop app.
Attackers have found a novel way to weaponize a legitimate feature of OpenAI's ChatGPT platform: the share link function. By creating fake outage notifications hosted on ChatGPT's own infrastructure, threat actors are luring users into downloading malware disguised as the ChatGPT desktop application. The campaign, reported by BleepingComputer, exploits user trust in the AI platform and the urgency created by a perceived service disruption.
The attack works by generating a ChatGPT share link that renders a page mimicking an official OpenAI outage notice. The page claims that ChatGPT is temporarily unavailable and prompts the user to download a 'desktop client' to restore access. In reality, the downloaded file is malware — likely an infostealer or remote access trojan — that compromises the victim's system. Because the page is hosted on ChatGPT's legitimate domain, it bypasses many traditional security filters that would flag a third-party site.
This technique is particularly insidious because it leverages the reputation of a trusted brand. Users who see a chat.openai.com URL may let down their guard, assuming the content is safe. The social engineering is further amplified by the fake outage narrative, which creates a sense of urgency and encourages hasty action. Security researchers warn that similar abuse of platform-native features is a growing trend, as attackers seek to exploit the implicit trust users place in major internet services.
The campaign targets both Windows and macOS users, with the malware payloads varying by platform. While the exact malware families have not been publicly identified, early analysis suggests the payloads are designed to steal credentials, cryptocurrency wallets, and other sensitive data. The use of ChatGPT's share link feature also makes takedown more difficult, as the malicious pages are hosted on OpenAI's own servers and may require the company's direct intervention to remove.
OpenAI has not yet issued a public statement regarding the abuse, but the company is reportedly investigating the campaign. Users are advised to be cautious of any unsolicited prompts to download software, even if they appear to come from a trusted source. The safest course of action during a real outage is to check OpenAI's official status page or social media channels rather than clicking on shared links.
This incident underscores a broader shift in cybercriminal tactics: rather than building fake websites from scratch, attackers are increasingly abusing the features of legitimate platforms — from Google Docs to GitHub to ChatGPT — to host malicious content. These 'living off the land' attacks are harder to detect and block because the infrastructure is inherently trusted. As AI platforms become more integrated into daily workflows, security experts expect to see more campaigns that weaponize their sharing and collaboration features.