The assumption that a curated app store or extension marketplace is a safe source of software is one of the most consequential security misconceptions in the current enterprise environment. Google’s Chrome Web Store has a review process, and that review process provides some protection against obviously malicious submissions. It does not provide the comprehensive vetting that the existence of a review process implies to most users, and the recent discovery by Socket cybersecurity researchers of harmful code hidden within extensions presenting themselves as ChatGPT productivity tools demonstrates precisely where that gap is being exploited. The extensions in question, published under the name “ChatGPT Mods,” do not deploy malware or attack systems directly. They exploit the authentication process of a trusted platform to steal session credentials, giving attackers authenticated access to ChatGPT accounts without the account owner’s knowledge. Any organization whose employees have installed extensions from this publisher should treat removal as an immediate priority.
How the Attack Works and Why It Is Effective
The mechanism these extensions use is worth understanding in detail because it illustrates why conventional security thinking, focused on malware detection and direct system attacks, does not reliably catch this category of threat.
Once installed, the extension injects a script into chatgpt.com that monitors all outbound requests from the browser. When a request contains authorization details or session token data, the script extracts that information and transmits it to an attacker-controlled server. The attacker then uses the stolen token to authenticate ChatGPT sessions under the victim’s identity, gaining access to the account’s chat history and any applications connected to the platform. The extension also transmits metadata, usage telemetry, and backend-issued access tokens, expanding the information available to the attacker beyond just the immediate session.
The reason this approach is effective is that it operates entirely within the permissions that browser extensions legitimately use. An extension that monitors web requests and interacts with page content is doing something that thousands of legitimate extensions do. The malicious behavior is not in the category of action the extension takes but in what it does with the information it collects. This makes detection through standard security tools difficult, because the activity that constitutes the attack does not look meaningfully different from the activity of a legitimate extension performing its stated function.
The consequence of a compromised ChatGPT account is more significant than it might initially appear for organizations whose employees use the platform for work. Chat histories may contain sensitive business information, client details, internal analysis, or proprietary content that employees have shared with the AI in the course of their work. Connected applications extend the access further. An attacker operating under a legitimate user’s authenticated identity has access to everything that user has shared with the platform and everything the platform can reach through connected integrations.
The Affiliate Hijacking Campaign Targeting E-Commerce Platforms
The ChatGPT credential theft extensions are not an isolated incident. Socket researchers have identified a separate but related campaign targeting e-commerce platforms through browser extensions that combine legitimate advertised functionality with hidden revenue fraud.
The most clearly documented example is the “Amazon Ads Blocker” extension published by “10Xprofit” in January 2026. The extension functions as described, blocking Amazon advertising content for users who install it, which provides the genuine utility needed to accumulate installs. Simultaneously, it replaces existing Amazon product affiliate codes with the developer’s own affiliate tag, redirecting the commission revenue that would otherwise go to legitimate affiliates to the extension’s publisher instead.
The scale of the campaign extends well beyond this single extension. Researchers have identified 28 additional browser add-ons operating as part of the same scheme, targeting affiliate programs across AliExpress, Amazon, Best Buy, Shopify, Shein, and Walmart. For organizations that operate affiliate programs or rely on affiliate revenue, the practical impact is financial: transactions that should be attributed to legitimate partners are being credited to fraudulent ones, quietly diverting revenue that the affiliate arrangement was intended to direct elsewhere.
The pattern these two campaigns share is meaningful for how organizations think about browser extension risk. In both cases, the extension delivers enough of its advertised functionality to justify installation and avoid obvious suspicion. The malicious behavior operates in parallel with legitimate behavior rather than replacing it, which means users who install the extension and find that it works as described have no obvious reason to investigate further. This design pattern, functional enough to be installed, deceptive enough to be missed, is increasingly the standard approach for browser extension threats.
What Organizational Browser Security Needs to Address
The response to this category of threat requires operating at multiple levels simultaneously, because no single control addresses all of the ways malicious extensions enter and persist in enterprise environments.
Extension policy is the foundational control, and the most effective version of it is a default-deny approach that permits only pre-approved tools rather than attempting to identify and block malicious ones after the fact. The volume and velocity of extension publishing across major browsers makes comprehensive review of everything employees might choose to install impractical. A policy that defines an approved list of extensions, establishes a clear process for requesting additions to that list, and restricts installation of unapproved extensions through browser management policy reduces the attack surface to a manageable set of tools that IT and security teams can actually monitor and assess. This approach requires investment in the approval process to ensure it is responsive enough that employees are not routinely bypassing it because getting a tool approved takes too long, but that investment is far smaller than the cost of managing incidents that result from unrestricted extension installation.
Automated browser security tools provide the monitoring layer that policy alone cannot. Specialized solutions that assess installed extensions in real time, flagging excessive permissions, unexpected data transmission behavior, or sudden changes in extension behavior following an update, identify risks that would not be visible through manual review. Excessive permissions are a particularly useful signal: an extension requesting access to all website data, clipboard content, or browser history when its stated function does not require that access is exhibiting the permission pattern that malicious extensions need to operate. Many legitimate extensions also request broader permissions than they strictly need, which is a separate problem, but the combination of broad permissions and a newly published or recently updated extension warrants closer examination.
Scheduled audits of installed extensions across the organization provide the visibility needed to identify unauthorized installations, extensions that are no longer in active use, and extensions that have changed behavior following updates. This last category is particularly important because an extension can be legitimate at the time of installation and become malicious following an update that introduces harmful code, either because the developer has made that choice deliberately or because the extension’s publisher account has been compromised by an attacker who then pushes a malicious update to the existing install base.
Employee education on browser extension risk needs to be specific enough to be actionable. General awareness that extensions can be malicious is insufficient when the extensions in question are delivering genuine functionality alongside their hidden behavior. The more useful training addresses the specific signals that warrant scrutiny: permission requests that exceed what the extension’s stated purpose requires, publishers with limited history or reviews, extensions that recently appeared and are claiming to enhance popular platforms, and the importance of reporting unusual account behavior that might indicate credential theft has occurred. Employees who understand that a working extension is not the same as a safe one, and who know what the specific risk indicators look like, are better positioned to make installation decisions that reflect the actual risk landscape.
The broader principle that these incidents reinforce is that the security perimeter for modern enterprise environments extends to every tool that employees install in their browsers, and that the curation provided by extension marketplaces is not a substitute for organizational oversight of what is actually running in those browsers. The extensions targeting ChatGPT credentials and affiliate revenue are effective precisely because they exploit the trust that users extend to tools installed through official channels. Managing that trust gap is an organizational responsibility, and the controls that address it are available and implementable for organizations that treat browser security as a component of their broader security posture rather than an afterthought.
