Fingerprint, a US-based device intelligence platform for fraud prevention, has announced the launch of its Model Context Protocol (MCP) Server, described as the first open-source MCP implementation in the fraud prevention sector.
The server enables organisations to connect any AI assistant, chatbot, or agent directly to Fingerprint's device intelligence platform, allowing fraud analysts to query data and investigate anomalies using natural language prompts without writing code.
The MCP Server connects to Fingerprint's production data via APIs, returning real-time insights from device intelligence events. Fraud teams can ask questions such as whether login attempts are connected or what patterns exist across suspicious sessions, with the AI assistant querying Fingerprint through the server and returning results immediately. The company states this reduces investigation cycles from hours to seconds.
Open standard and deployment options
By adopting the Model Context Protocol, an open standard rather than a proprietary integration layer, Fingerprint allows organisations to use their preferred AI tools, including custom agents or commercial assistants such as Claude or ChatGPT, without vendor lock-in. The server will be available as both open-source software and a managed service, and supports integrations with third-party tools and systems.
Beyond read-only data access, the MCP Server connects AI agents to Fingerprint's Management API, enabling configuration and management of fraud prevention workflows. Developers can use the server to build AI-powered fraud investigation tools, automated incident monitoring workflows, and fraud-aware applications using AI coding environments such as Claude Code or Cursor.
The launch follows Fingerprint's Authorised AI Agent Detection ecosystem and forms part of a broader agentic AI roadmap, with additional capabilities planned throughout 2026. The MCP Server is currently available on an invitation-only basis to select enterprise organisations. Model Context Protocol downloads grew from 100,000 in November 2024 to over eight million by April 2025, reflecting the rapid adoption of the standard across industries.
Commenting on the news, Valentin Vasilyev, Co-founder and CTO at Fingerprint, said the MCP Server gives fraud teams the freedom to use the AI tools they already trust
