Analysis · curated 28 Jul 2026

AI Agent Access Control: Why 91% of MCP Servers Skip OAuth

First reported today guild.ai

Coverage timeline

Single-source analysis — first reported, latest, and curated coincide.

Why it matters

MCP servers authenticating agents with long-lived static credentials outside SSO create a broad exposure surface for credential theft and agent abuse that traditional IAM and shift-left scanning do not catch.

An analysis by Guild.ai synthesizing 2026 survey data reports that 91.5% of MCP servers in the official registry rely on static API keys, personal access tokens, or no authentication, with only 8.5% using OAuth, and that 42,000 MCP servers were exposed to the public internet in Q1 2026 leaking API keys, Slack credentials, and chat histories. The piece argues that AI agent access control is a credential-issuance and scoping problem, not a role problem, and offers permission-management guidance for platform teams.