Material maps OAuth-centric Google Workspace attack chain and AI agent risk

Material has published an analysis of what it calls the modern workspace attack chain, arguing that recent incidents at Vercel and Composio were the same attack run twice against different targets, with OAuth grants rather than email as the entry point into Google Workspace.

The company said attackers use an OAuth token to reach email and Drive, read sensitive data and then pivot beyond the workspace, and that AI agents connected to Workspace through legitimate OAuth grants can follow the same path without any malicious actor or compromised credential involved.

Material’s coverage spans email payload blocking, activity-level detection of suspicious OAuth behaviour, visibility into sensitive data at rest across email and Drive, and redaction of password reset links with step-up verification to block lateral movement.