Malicious Chrome extensions impersonate AI assistants and steal user data
Researchers discovered approximately 30 malicious Chrome extensions impersonating popular AI assistants that collectively infected between 260,000 and 300,000 users. The extensions, distributed through the Chrome Web Store and sharing identical backend infrastructure, granted attackers broad remote access to user browsers to steal credentials and monitor behavior.
What happened
Threat actors created approximately 30 malicious Chrome extensions that impersonate well-known AI assistants including ChatGPT, Gemini, Grok, Claude, and 'AI Gmail'. These fake extensions collectively received over 260,000 to 300,000 installations.
The extensions shared the same internal structure, logic, permissions, and backend infrastructure. Once installed, they granted attackers wide-ranging remote access to the user's browser, including voice recognition, pixel tracking, and the ability to read email content through injected iframes. Some extensions were even marked as 'Featured' in the Chrome Web Store.
Reported impact
Security researchers identified at least 260,000 to 300,000 users who installed these extensions. The malicious extensions enabled attackers to harvest user data and monitor user behavior by gaining remote control over browser functionality.
Index analysis
AI-powered browser extensions have become a popular vector for threat actors seeking to harvest user information. The campaign, dubbed AiFrame, demonstrates how attackers exploit user trust in familiar AI brand names to distribute malware at scale.
Governance controls
The extensions were distributed through the official Chrome Web Store, where some were marked as 'Featured' content, increasing their apparent legitimacy and download numbers. This suggests gaps in Chrome Web Store review processes for detecting impersonation and malicious permissions.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.