As reported by BleepingComputer, a fraudulent Chrome extension impersonating the popular Perplexity AI platform was discovered actively harvesting user search queries and browsing data from the Chrome Web Store. While the mechanics are familiar, the broader implications for enterprise security and end-user trust deserve careful examination.
Why This Matters More Than It Appears
Browser extensions remain one of the most underestimated attack surfaces in modern security programs. Unlike traditional malware, extensions operate with elevated browser-level permissions — often with explicit user consent — making them uniquely positioned to intercept sensitive data streams without triggering traditional endpoint detection controls. The use of a recognizable, trusted AI brand like Perplexity is not incidental; it is the core of the attack strategy. As AI tools have rapidly entered mainstream workflows, users are actively seeking extensions that enhance their AI experience, creating a fertile environment for this kind of brand impersonation.
Search query interception is particularly dangerous because it captures intent — not just what sites a user visits, but what they are actively investigating, researching, or considering. For enterprise users, this can expose:
The Chrome Web Store Vetting Problem
This incident reinforces a persistent structural weakness: Google's Chrome Web Store review process remains insufficient to consistently catch malicious extensions before they reach users at scale. Despite repeated high-profile cases of malicious extensions slipping through, the vetting pipeline has not evolved fast enough to match the volume and sophistication of submissions. Threat actors have learned to publish extensions that appear benign at submission time, with malicious behavior activated post-approval or triggered only under specific conditions.
Shield53 Perspective: The Chrome Web Store should not be treated as a trusted marketplace equivalent to a curated enterprise software catalog. The presence of an extension in the store is a necessary but entirely insufficient indicator of safety.
Who Is Most at Risk
This threat disproportionately affects several groups:
- Knowledge workers and researchers who rely heavily on search and AI tools throughout their day
- Organizations without browser extension governance policies, where employees freely install extensions on corporate devices or profiles
- Early AI adopters who eagerly install AI-branded tools without scrutiny
- Remote and hybrid workers using personal browsers or unmanaged devices for work tasks
Shield53 Recommendations
For Security Teams and CISOs
- Implement browser extension allowlisting via Chrome Enterprise or equivalent MDM controls. Only pre-approved extensions should be installable on managed devices. This single control would have entirely prevented this specific attack vector.
- Audit installed extensions across your fleet immediately. Use tools like Google Admin Console, Tanium, or CrowdStrike to enumerate all browser extensions and flag any that are AI-branded, newly published, or have excessive permissions.
- Monitor extension permissions at onboarding. Any extension requesting access to browsing history, search data, or "read and change all your data on websites you visit" should require explicit security team approval.
- Subscribe to threat intelligence feeds that specifically track malicious browser extensions. Organizations like SANS ISC and vendors like SquareX or LayerX specialize in this space.
For End Users
- Always verify extensions by navigating directly to an AI tool's official website to find their legitimate extension link — do not search the Chrome Web Store for AI tools by name.
- Review the publisher name, publication date, and user review count critically. Legitimate tools from major AI companies will have verifiable publisher identities and substantial install bases.
- Treat any extension requesting broad browsing permissions with significant skepticism, regardless of how trusted the brand appears.
Immediate Actions
- Search your Chrome Web Store admin dashboard or endpoint telemetry for any extension with "Perplexity" in its name not published by the verified Perplexity AI developer account.
- If found, remove immediately, rotate any credentials that may have been searched for during the exposure window, and consider a broader password hygiene review for affected users.
The broader lesson here extends well beyond this single extension: the rush to adopt AI-branded tooling has outpaced the security hygiene needed to safely integrate these tools. Until browser vendors enforce more rigorous publisher verification and behavioral analysis at scale, the extension ecosystem will continue to be exploited by actors who understand that user enthusiasm is a vulnerability in itself.
