Teleport launches Secure MCP to protect AI enterprise workflows
Teleport has announced the general availability of its Secure Model Context Protocol (MCP) for use on the Teleport Infrastructure Identity Platform.
The Secure MCP solution seeks to address new security challenges emerging from the rapid adoption of artificial intelligence across enterprises. Recent data from Enterprise Strategy Group indicate that 44% of enterprises have now deployed AI within their organisations. Teleport's Secure MCP is designed to provide security guardrails for AI systems as they interact with databases, MCP servers, and other forms of enterprise data.
The Model Context Protocol is an open standard that enables AI models to connect with various tools, databases, or applications using a simplified, universal interface. This is intended to streamline integration in a manner akin to technology standards such as USB-C for physical devices. Despite these integration benefits, MCP was not originally intended with access control, which presents risks around unrestricted data access for AI models. Consequently, there is a need for mechanisms that can provide controlled, audited, and secure access to sensitive data.
Teleport's Secure MCP responds to these needs by employing its Infrastructure Identity Platform, which extends existing trust frameworks to AI-based workflows. The platform enforces both Role-Based and Attribute-Based Access Controls (RBAC and ABAC) to manage the resources that large language models (LLMs) can access. Every session involving AI data access is logged, thereby contributing to regulatory compliance and audit readiness.
Ev Kontsevoy, Chief Executive Officer of Teleport, commented on the development:
"AI is terraforming how software is deployed in organizations. It shouldn't require a major public security incident to motivate business leaders to prepare for this impending challenge. Applying the same access control guardrails for AI, humans, and non-human identities accelerates AI adoption while locking in the protection needed to prevent unauthorized access of data. That's why we launched our secure MCP solution for Teleport, to enable enterprises to confidently unlock AI's innovation without falling prey to its security vulnerabilities and loopholes."
Industry analysts have noted a concurrent rise in deployments of AI agents that operate within core enterprise systems, increasing the urgency for businesses to address identity and data security concerns. Todd Thiemann, Principal Analyst for Identity Security & Data Security at Enterprise Strategy Group, highlighted the pressing nature of these issues:
"A wave of AI agent deployments that touch on core enterprise systems is in process, and identity teams need to be prepared. Recent Enterprise Strategy Group research showed that data privacy and security for AI agents were major concerns for enterprise security teams. Teleport's Secure MCP solution lays the groundwork for secure agent deployment and enables identity teams to get ahead of the game in securing their AI agent deployments."
Secure MCP delivers several key architectural components for AI and MCP deployments. These include Zero Trust Networking, allowing only authenticated clients to interact with MCP servers over encrypted connections. A live MCP server inventory feature allows administrators to discover and register MCP tools across hybrid infrastructure environments automatically.
Strict access control ensures that language models are only able to access resources for which they are specifically authorised, while the principle of least privilege means that authorisations are granted on a just-in-time basis for defined tasks. This minimises the potential risk of overprivileged or persistent access by AI models. Additionally, comprehensive audit trails provide a record of every attempt - successful or denied - by LLMs to access data.
The extension of these security controls to MCP allows engineering teams to develop technology that incorporates AI without opening new avenues for unauthorised access to company data. By supporting both machine and user-driven LLM workflows, Teleport states its platform is positioned to accommodate a range of AI integration scenarios while maintaining a strong security posture.
