Shadow AI detection: see every harness, model, MCP server, and skill on your company's devices

Speakeasy has introduced Shadow AI detection for organizations using its AI Control Plane device agent, aiming to identify AI tools deployed outside formal procurement or security review. The feature scans enrolled macOS devices for four categories of shadow AI: coding harnesses and assistants, local or external model providers, MCP servers, and agent skills, then presents usage by tool, user, device, version, and last-seen time in a centralized dashboard. Administrators can mark tools or MCP servers as approved or blocked, with gateway enforcement preventing supported identified clients from accessing company MCP servers while leaving software removal to MDM or EDR systems. Detection uses a maintained library of roughly 40 scan targets, supports custom targets, runs matching locally on devices, and reports only discovered matches and scan records rather than prompts, shell history, or usage data. The capability is currently in beta for eligible organizations, with shadow AI scanning available on macOS and broader device-agent features available on Windows and Linux.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.