While much focus in AI security is on protecting ‘shadow’ AI and GenAI use, a significant vulnerability remains unaddressed: AI browser extensions.

A new report from LayerX reveals the extent of this oversight, suggesting AI extensions may be the most overlooked and dangerous AI threat within your network.

AI browser extensions bypass traditional security measures like DLP and SaaS logs, operating directly within the browser. They have direct access to employee activity. These plugins are significantly more likely to contain vulnerabilities and have greater access to sensitive data, yet install quickly and can persist indefinitely.

The common misunderstanding is that extensions are a minor risk, limited to specific users or situations, an incorrect assumption.

The report indicates almost all business users have at least one browser extension installed, with many having over 10. This is widespread.

However, many organizations can’t answer basic questions about extension usage, access, permissions, and data access.

Despite advancements in securing networks, endpoints, and identities, browser extensions remain a major security gap.

While current AI security discussions prioritize SaaS and APIs, this report calls attention to a different, largely ignored route: AI browser extensions.

These tools are rapidly gaining popularity. Roughly 1 in 6 enterprise users already utilize AI extensions with increasing adoption.

Companies may restrict or oversee access to AI applications; however, extensions operate differently. They live within the browser and can access content, inputs, and session data without being flagged by standard security measures.

Effectively, these extensions establish an uncontrolled space for AI use, bypassing visibility and policy controls.

It might be assumed that AI extensions have risks similar to regular extensions, the evidence suggests otherwise.

AI extensions pose a greater threat. They’re more prone to vulnerabilities, more likely to access cookies, have scripting permissions, and manipulate browser functions.

Each capability carries significant implications. Accessing cookies can expose session credentials. Scripting allows for data theft and manipulation. Tab control enables redirection for phishing or silent attacks.

Rapid adoption, heightened access, and minimal oversight, render AI extensions an important, new area for potential security compromise.

Security teams often view extensions as unchanging, but that is not accurate.

Extensions are constantly changing through updates, ownership changes, and permission expansions.

The report demonstrates AI extensions are far more likely to alter their permissions over time. Most users have at least one AI plugin with changing permissions.

This situation poses a consistent challenge that traditional allowlists struggle to address. Safe extensions might not be safe going forward.

Security teams employ various trust indicators to assess extensions, including developer transparency, install numbers, update frequency, and privacy policies. While not directly indicating malicious behavior, these are key indicators for evaluating overall risk.

A large number of extensions have very small user bases. Many extensions reach only a few thousand users. AI extensions are particularly concerning for this risk profile. Limited adoption poses heightened risk to the application.

Additionally, many extensions have not been updated in over a year, meaning the developer may no longer be actively engaged. Outdated extensions are prone to vulnerabilities that attackers can exploit.

Consequently, most extensions within corporate environments show weak or missing signals within these key assessment areas. This raises serious questions regarding data handling and regulatory compliance. Further, this reinforces the fact that extensions are less scrutinized than other software components.

The report defines a clear plan for security teams:

For a long time, browser extensions have been seen as a feature of convenience that enable productivity and customization. However, they now are a significant part of the overall attack surface. Because they are popular, have elevated permissions, and go largely unmonitored, they represent direct exposure to private data and user sessions.

Download the complete Extension Security report from LayerX to get the full details of the findings, identify where your potential risk lies, and gain a clear plan to manage this emerging attack surface without sacrificing productivity.