A malicious Chromium extension masquerading as a Perplexity AI tool leveraged Manifest V3 (MV3) APIs to intercept and log real-time address bar keystrokes before user submission. By implementing a redirection pattern (User $\rightarrow$ Attacker Intermediary $\rightarrow$ Legitimate Search Provider), the threat actor captured sensitive queries, PII, and credentials without disrupting the user experience. This human-layer attack highlights a critical governance gap in browser extension auditing, allowing for silent reconnaissance and intellectual property theft within corporate environments via attacker-controlled intermediary infrastructure.

• Attack Vector: AI-Brand Impersonation

  • Leveraged the rapid adoption and trust of Perplexity AI to conduct "human-layer" social engineering.
  • Distributed as a Chromium extension requiring broad website access permissions to facilitate data interception.
  • Designed to maintain a functional browsing experience to avoid triggering user suspicion or immediate removal.

• Technical Mechanics: MV3 Interception

  • Utilized Manifest V3 (MV3) APIs specifically designed for search redirection and browser interception.
  • Implemented real-time keystroke logging in the address bar, capturing data before the user pressed the "Enter" key.
  • Routed all intercepted traffic through attacker-controlled intermediary servers before forwarding the request to the legitimate search provider.

• Impact & Operational Risk

  • Data Compromise: Targeted PII, corporate credentials, and sensitive business intellectual property contained within search intent.
  • High Stealth: The transparent nature of the redirection ensured the user observed no latency or functional errors.
  • Reconnaissance Value: Provided attackers with a persistent window into employee workflows and internal corporate queries.

• Enterprise Governance & Strategic Response

  • Highlighted a systemic blind spot where organizations possess mature software inventories but lack visibility into browser extensions.
  • Necessitates a shift toward secure enterprise browser technologies to enforce centralized extension auditing and allow-listing.
  • Gartner predicts 30% of enterprises will adopt secure browser frameworks by 2029 to mitigate these extension-based risks.

• Conclusion & Remediation

  • The attack underscores the evolution of AI-themed lures to bypass traditional security awareness training.
  • Immediate defensive actions include auditing installed browser extensions for unauthorized MV3 permissions.
  • Long-term strategy requires integrating browser extension telemetry into broader XDR/SIEM monitoring frameworks.