Fake Perplexity extension on Chrome Web Store tracked searches

# Fake Perplexity Extension on Chrome Web Store Hijacked Search Queries, Tracked User Browsing

A malicious Chrome extension masquerading as Perplexity AI was discovered on Google's Chrome Web Store, intercepting search queries and collecting browsing data from unsuspecting users. Microsoft Threat Intelligence identified the impostor extension, which used deceptive branding and routing mechanisms to surveil user search behavior while forwarding queries through attacker-controlled infrastructure.

## The Threat: Search Hijacking at Scale

Security researchers at Microsoft identified an extension titled "Search for perplexity ai" (extension ID: flkebkiofojicogddingbdmcmkpbplcd) that posed as the legitimate Perplexity AI research assistant. The extension was active on the Chrome Web Store—Google's official marketplace for browser extensions—where it was available for download by any Chrome user.

Once installed, the malicious extension:

The extension used a counterfeit domain (perplexity-ai[.]online) to establish authority, distinguishing itself only slightly from the legitimate Perplexity service at perplexity.ai. This subtle difference in branding made the extension appear authentic to users unfamiliar with the official product.

## Background and Context: Why Perplexity Is a Target

Perplexity AI has emerged as a significant player in the conversational search market, offering users an alternative to traditional search engines. Unlike Google or Bing, Perplexity synthesizes real-time web information into direct, conversational responses rather than presenting ranked link results. The service is available across multiple platforms:

The legitimacy and growing adoption of Perplexity's services made it an attractive target for threat actors. Users actively searching for an "official" Perplexity extension might install the impostor without verification, providing immediate access to their search behaviors and browsing patterns.

## Technical Details: How the Malicious Extension Operated

### Search Interception Mechanism

The fake extension intercepted user searches at the browser level using Chrome's chrome_settings_overrides manifest permission. This allowed the extension to:

1. Replace Chrome's default search provider with the attacker's search intermediary

2. Capture every query typed into the address bar before processing

3. Log search data on the attacker's infrastructure with timestamp and context

4. Redirect the user to legitimate search results after data collection

Microsoft researchers found intentional logging code on the attacker's server infrastructure, confirming this was deliberate surveillance rather than an accidental privacy leak.

### Permission Escalation Risk

The extension requested several powerful permissions that exceeded typical AI assistant functionality:

| Permission | Legitimate Use | Risk Factor |

|-----------|------------------|------------|

| declarativeNetRequest | Traffic filtering | Enable credential interception and password harvesting |

| webRequest | URL monitoring | Inspect encrypted traffic; block sensitive domains |

| activeTab | Access page content | Read sensitive form data and account credentials |

| <all_urls> | Universal site access | Monitor all user browsing activity |

"The extension requests powerful DNR permissions that enable traffic redirection, URL rewriting, and selective request filtering, which aren't consistent with expected AI assistant behavior," Microsoft noted in their analysis.

While the current version did not actively harvest credentials, the permission configuration would allow rapid extension of the malware's capabilities with minimal modification.

### Data Collection Scope

The intercepted data included:

This level of profiling could enable targeted phishing campaigns, account compromise, or corporate espionage if users searched for sensitive information.

## Implications for Users and Organizations

### Individual Risk

Users who installed the extension faced several risks:

### Enterprise Risk

For organizations, the implications were more severe:

## Why This Matters: The Chrome Web Store Trust Problem

This incident highlights a persistent vulnerability in Google's own extension marketplace. Despite controlling the Chrome Web Store, Google faces significant challenges in detecting malicious extensions before they reach users:

1. Scale of submissions: Thousands of extensions upload daily, making manual review impractical

2. Deceptive tactics: Threat actors use legitimate-looking names, icons, and branding to evade automated detection

3. Permission obfuscation: Extensions can request capabilities that seem reasonable initially but enable broader attacks

4. Installation momentum: Once an extension reaches users, it can persist silently for months

This case demonstrates that no official app store is immune to supply chain compromise—a critical lesson as organizations depend on browser extensions for work productivity and authentication.

## Recommendations: Immediate and Long-Term Actions

### For Individual Users

### For Enterprise Security Teams

### For Browser Vendors and App Stores

---

## HackWire Analysis

This attack represents a maturation in supply chain targeting tactics. Rather than compromising the software vendor itself, threat actors are poisoning the distribution channel—exploiting the trust users place in official marketplaces. What makes this particularly effective is the permission architecture gap: Chrome extensions are powerful enough to intercept all user traffic, yet the review process treats them as low-risk add-ons.

The timing is significant. As AI tools like Perplexity gain adoption, they become attractive attack surfaces. Users actively seeking AI research assistants are likely to be tech-forward (and research-oriented), making them valuable targets for profiling and eventual compromise. A threat actor who can see what security researchers, engineers, or business strategists are searching for has gained asymmetric intelligence.

The critical insight: Microsoft noted the extension did not steal credentials, but had permissions to do so. This suggests the attacker was in the data collection phase, building profiles for future monetization or espionage. This is the opposite of what most users assume—that malware steals immediately. Instead, sophisticated attackers collect leverage first, then weaponize later.

This also exposes a friction point in Google's own ecosystem. Google controls Chrome, the Chrome Web Store, and could theoretically verify extensions cryptographically. That they haven't done so at scale suggests either technical debt or prioritization misalignment. Until browser vendors treat extension signing with the same rigor as OS-level code, users will remain vulnerable to supply chain attacks through what feels like the most official distribution channel.

For defenders: This incident is a reminder that network visibility matters more than ever. A proxy or DNS filter that logs all traffic to unknown domains would have immediately flagged perplexity-ai[.]online. For users: Never assume an app store badge means safety—always verify with the official vendor directly.

— HackWire Editorial

---

## Related Coverage