Fleet has launched AI governance features within its device management platform to give IT and security teams oversight of AI tools on employee devices across Mac, Windows, and Linux.

Session context visibility shows what users send to AI tools and how those tools respond, aiding security and compliance teams in detecting data sharing of corporate data, customer information, or source code.

New functions address employees installing AI agents, browser extensions, local models, plugins, and MCP servers outside formal procurement or approval processes.

AI inventory and visibility are available in Fleet Free, allowing users to view installed AI tools, agents, skills, plugins, and MCP servers with filters by publisher, version, and device.

Token usage tracking by tool, user, and device enables visibility into how AI applications are used and where spending occurs across managed devices.

Features include AI inventory, token usage and spending tracking, session visibility, governance controls, and auditable workflows, all integrated into Fleet's existing platform.

Fleet offers an open-source agent to provide transparency about data collection and tool usage as part of its AI oversight strategy on managed devices.

Mike McNeil, CEO and Co-Founder of Fleet, emphasizes the lack of visibility into AI software on staff laptops and the problem of tool sprawl without a unified governance approach.

Fleet argues that shadow AI should be governed in the same place as plugins, extensions, software, and developer tools, with device-level governance as the starting point.

Fleet frames the launch as addressing security concerns from SaaS sprawl to unmanaged AI use on endpoints, noting employees often adopt AI tools faster than central IT can assess them.

The launch highlights endpoint governance as a growing area, since employees can install tools locally with limited central visibility, including coding assistants, desktop utilities, browser add-ons, and external-model connections.

Governance controls align with Fleet's existing policy, reporting, and remediation workflows and can be managed through infrastructure-as-code and GitOps for reviewable and reversible changes across operating systems.