Firefox Enterprise: Self-Hosted Controls for Shadow AI

Firefox Enterprise is Mozilla's new managed browser platform, and its core idea is that the organization, not the vendor, holds the controls. Mozilla announced the product on October 12, 2026. It pairs the Firefox browser with a management console that a company can run on its own infrastructure, and it includes specific controls for how staff use AI services at work.

• What it is: Firefox plus a management console that organizations can self-host.
• AI controls: allow, limit or block AI services, per user or per group.
• Data controls: block or warn on paste, upload, download and print on selected sites.
• Availability: an initial sales focus on France, Germany, Austria and Switzerland.

What Is Firefox Enterprise?

The browser is where most work now happens, and that makes it the place where most data leaves. Firefox Enterprise treats the browser as a control point.

According to Mozilla, IT teams use the console to manage browser policy and access centrally, control how data moves through the browser, and set rules around AI use. It is designed to work alongside systems a company already runs, including identity providers, SIEM platforms and device management tools.

Help Net Security's report adds the detail that matters to a security team. Because the console is self-hosted, the organization controls its own management data, encryption keys, telemetry and policies. Connections to Mozilla and third-party services that a managed deployment does not need are disabled.

How Does Firefox Enterprise Control Shadow AI?

Shadow AI is the use of AI tools that an organization has not approved. The usual pattern is simple: an employee pastes a customer list or a block of source code into a chatbot to save time. There is no malice, and the data has still left the building.

Firefox Enterprise addresses this in two layers, as described by Help Net Security.

• AI controls: administrators can allow, limit or block AI services, and can turn off the browser's own AI features. Policies apply per user or per group, so a research team and a finance team can have different rules.
• Data-sharing controls: administrators can block or warn when someone pastes text, uploads or downloads files, or prints on selected websites. The checks run on the device, or a compatible third-party system can handle them.

The second layer is the stronger one. Blocking a list of sites is a game of catch-up, because new AI services appear every week. Controlling the action itself, the paste or the upload, works regardless of which site is on the other end. We covered a consumer version of the same idea in Opera Paste Protect.

Other Security Features in the Platform

Help Net Security lists several more controls:

• Device checks: sign-in can be denied when a device fails requirements such as disk encryption, and administrators can remotely sign out lost or stolen devices.
• Access Connector: reaches internal sites and apps through a company-controlled proxy, without a device-level VPN.
• Lockdown Mode: adds restrictions to selected sites and can isolate their browsing data.
• Encryption: stored history, logins and other profile data are encrypted by default.
• Watermarks: identifying marks on sensitive pages help trace shared screenshots.
• Logging: browser events and administrator actions can be exported to a security monitoring system.

The platform also integrates with existing data protection tools from Trellix and Broadcom.

Who Can Get It?

Mozilla says the rollout begins in Europe, with an initial sales focus on France, Germany, Austria and Switzerland and more markets to follow. Firefox Enterprise is a paid product offered through a demo request, and pricing has not been published. The standard Firefox browser and its Extended Support Release remain free.

Mozilla also says a Thunderbird Enterprise product is in development, extending the same customer-controlled approach to email.

The Takeaway for Security Teams

Three things make this worth evaluating. Policy is enforced where the data moves. The management plane runs on infrastructure you control. And AI use becomes a setting you can tune per team instead of a blanket ban that people work around.

It also fits a wider pattern of platforms adding guardrails for AI at the point of use, like Apple's full disk access controls for AI agents on macOS. If your organization is still writing its AI usage policy, a browser that can enforce one is a practical place to start. More defensive tooling is in our AI security coverage.

Sources: Mozilla blog, Introducing Firefox Enterprise — October 12, 2026; Help Net Security — October 13, 2026.