Google has always struggled with the issues of extensions in Chrome. In June 2023, the company had to
manually remove 32 exploitable extensions that were installed 72 million times before they were taken down.
Google’s previous extension framework, Manifest Version 2 (MV2), was notoriously problematic. It often granted excessive permissions to extensions and allowed scripts to be injected without user awareness, making it easier for attackers to steal data, access sensitive information, and introduce malware.
In response, Google introduced Manifest V3, which aimed to tighten security by limiting permissions and requiring extensions to declare their scripts in advance. While MV3 was expected to resolve the vulnerabilities present in MV2, SquareX’s research shows that it falls short in critical areas.
Malicious extensions built on MV3 can still bypass security features and steal live video streams from
collaboration platforms like Google Meet and Zoom Web without needing special permissions. They can also add unauthorized collaborators to private GitHub repositories, and even redirect users to phishing pages disguised as
password managers.
Furthermore, these malicious extensions can access browsing history, cookies, bookmarks, and download history, in a similar way to their MV2 counterparts, by inserting a fake software update pop-up that tricks users into downloading the malware.
Once the malicious extension is installed, individuals and enterprises cannot detect the activities of these extensions, leaving them exposed. Security solutions like
endpoint protection, Secure Access Service Edge (SASE), and
Secure Web Gateways (SWG) cannot dynamically assess browser extensions for potential risks.
