By Ava Callegari
Publication Date: 2025-11-24 20:20:00
Perplexity’s Comet Conundrum: Denials, Patches, and the Shadows of AI Browser Security
In the rapidly evolving world of AI-driven web browsers, Perplexity’s Comet has emerged as a bold entrant, promising to revolutionize how users interact with the internet through intelligent agents. But recent allegations of severe security vulnerabilities have cast a long shadow over its ambitions. Security researchers from SquareX claimed to have uncovered a hidden API in Comet that could allow attackers to execute arbitrary code on users’ devices, potentially leading to data theft or full system compromise. Perplexity, however, has vehemently denied these claims, labeling them as “fake news” and insisting that no such vulnerability ever existed.
The controversy erupted when SquareX published a detailed report outlining what they dubbed a “major security flaw” in Comet’s architecture. According to their findings, the browser’s Multi-Context Protocol (MCP) API, intended for internal use, was allegedly accessible externally, enabling malicious actors to inject commands that could manipulate browser behavior or access sensitive user data. This isn’t the first time Comet has faced scrutiny; earlier reports from Brave and Guardio highlighted issues like prompt injection vulnerabilities, where hidden instructions on webpages could trick the AI into performing unintended actions, such as closing tabs or opening phishing sites.
Perplexity’s response was swift…