AI Coding Agents Leak Credentials via MCP Config Files, Exposing Organizations to Cyberattacks

Thumbnail Image

The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.

AI coding agents using the Model Context Protocol (MCP) have led to widespread credential leaks through insecurely stored configuration files. Research found thousands of valid secrets, including API keys and passwords, exposed on public repositories, resulting in unauthorized access and financial losses for organizations. Security experts warn of significant risks as MCP adoption grows.[AI generated]

Why's our monitor labelling this an incident or hazard?

The event involves AI systems because MCP config files are used to connect AI coding agents to enterprise tools and data, enabling autonomous AI operation. The hardcoded credentials in these files have been exploited by attackers, leading to direct harm including unauthorized access to sensitive corporate systems and financial losses from AI token misuse. This fits the definition of an AI Incident as the AI system's use has directly led to harm to property (corporate data and networks) and financial harm (costly AI token bills).[AI generated]
AI principles
Privacy & data governanceRobustness & digital security

Industries
Digital security

Affected stakeholders
Business

Harm types
Economic/Property

AI system task:
Content generation


Articles about this incident or hazard