Give Them an Inch and They Will Take a Mile:Understanding and Measuring Caller Identity Confusion in MCP-Based AI Systems

TL;DR

This paper investigates security vulnerabilities in MCP-based AI systems, revealing that lack of caller authentication and coarse authorization mechanisms expose these systems to significant risks, emphasizing the need for improved security protocols.

Contribution

The study provides a comprehensive security analysis of MCP servers, highlighting critical vulnerabilities and proposing the necessity for explicit caller authentication and fine-grained authorization.

Findings

Most MCP servers rely on persistent authorization states.

Many MCP servers do not enforce per-tool authentication.

Lack of caller authentication increases attack surface.

Abstract

The Model Context Protocol (MCP) is an open and standardized interface that enables large language models (LLMs) to interact with external tools and services, and is increasingly adopted by AI agents. However, the security of MCP-based systems remains largely unexplored.In this work, we conduct a large-scale security analysis of MCP servers integrated within MCP clients. We show that treating MCP servers as trusted entities without authenticating the caller identity is fundamentally insecure. Since MCP servers often cannot distinguish who is invoking a request, a single authorization decision may implicitly grant access to multiple, potentially untrusted callers.Our empirical study reveals that most MCP servers rely on persistent authorization states, allowing tool invocations after an initial authorization without re-authentication, regardless of the caller. In addition, many MCP servers fail to enforce authentication at the per-tool level, enabling unauthorized access to sensitive operations.These findings demonstrate that one-time authorization and server-level trust significantly expand the attack surface of MCP-based systems, highlighting the need for explicit caller authentication and fine-grained authorization mechanisms.