This AI agents 2026 readiness check covers the hard switch Google is flipping: starting November 11, 2026, Workspace MCP servers reject any OAuth client not flagged as agentic. Run the free check below — 3 questions tell you if you're affected, and the fix takes two minutes in the Cloud console.
Key figure: Multi-factor authentication reduced the risk of account compromise by 99.22% across Microsoft Entra ID sign-ins, and by 98.56% even when credentials had already leaked (Microsoft, 2023).
The deadline is November 11, 2026: Google Workspace MCP servers start rejecting OAuth clients not flagged as agentic. If your AI agents call Gmail, Drive, Docs or other Workspace services through MCP, tick “Use this client for an AI-powered agent” in Google Cloud console before the date. Miss the November 12 lock-in and the flag becomes permanent — an unflagged client can never be converted later.
Until Nov 11, 2026 00:00 UTC — when non-agentic clients start getting rejected.
1. Does an AI agent (yours, or one you operate) call Google Workspace — Gmail, Drive, Docs, Sheets, Slides, Calendar, Chat or People — through MCP?
2. Does the same OAuth client serve both human users (REST APIs) and agent workflows (MCP)?
3. Have you already ticked "Use this client for an AI-powered agent" on the agent's OAuth client?
gmailmcp.googleapis.com) at the flagged client and confirm it connects. Do this before Nov 11 — on Nov 12 the flag locks permanently on existing clients.This is the part most teams will miss. The toggle has been editable since October 12, 2026. Enforcement starts November 11. But on November 12, the agentic setting becomes permanent on existing clients — you can no longer flip it. An unflagged client is stuck non-agentic forever: to use Workspace MCP after that date you must mint a brand-new OAuth Client ID, mark it agentic, and update every config that referenced the old one (think mcp_config.json files scattered across machines). The cheap fix is ticking the box now; the expensive fix is a client-ID migration under pressure.
Apps and integrations that only call Google Workspace REST APIs — classic Gmail/Drive/Calendar API integrations with no agent and no MCP — are untouched by this change. The requirement applies only to OAuth clients calling the Workspace MCP servers (Gmail, Drive, Docs, Sheets, Slides, Calendar, Chat, People MCP endpoints).
Honest scope: an informational aid, not official Google guidance and not legal advice. Dates and mechanics follow Google's Workspace developer release notes (Oct 2026); Google can change them. Verify in your own Cloud console before the deadline.
Sources: Google Workspace developer release notes (Oct 2026) — "OAuth 2.0 clients calling the Google Docs/Drive MCP server are required to have 'Use this client for an AI-powered agent' selected"; AppsScriptPulse coverage of the Nov 11 / Nov 12 enforcement and lock-in dates.
Update, October 22, 2026: the same week, Microsoft attributed the ConnectWise Cloud Backup supply-chain compromise to the China-linked actor Storm-0940, whose AI agents used stolen cloud application credentials to enter customer tenants through apps those tenants already trusted, with no user sign-in at all. Agentic OAuth cuts both ways: flag your own agent clients before this deadline, and audit every application already trusted in your tenant. Run the ARTEX Agent App-Credential Check to score one granted app at a time.
What happens on November 11, 2026 to Google Workspace MCP?
Google Workspace MCP servers (Gmail, Drive, Docs, Sheets, Slides, Calendar, Chat, People) start rejecting requests from any OAuth client not marked as agentic. The toggle lives in Google Cloud console → Google Auth Platform → Clients → "Use this client for an AI-powered agent".
What happens if I miss the November 11 deadline?
On November 12, 2026 the agentic setting locks permanently on existing OAuth clients. If yours isn't flagged, you can't flip it later — you must create a brand-new OAuth Client ID, mark it agentic, and update every config referencing the old client.
Are non-agent apps affected?
No. Apps that only call Google Workspace REST APIs are not affected — the requirement applies only to OAuth clients calling the Workspace MCP servers.
Can one OAuth client serve both humans and an AI agent?
Google says to keep them separate: one Client ID for human-interactive traffic (REST APIs), another for agent workflows (MCP). Mark only the agent's client as agentic. The flag grants no extra scopes and doesn't bypass user consent.
Where is the agentic toggle?
Google Cloud console → Google Auth Platform → Clients → select your OAuth client → check "Use this client for an AI-powered agent". Editable since October 12, 2026.
Starting November 11, 2026, Google Workspace MCP servers (Gmail, Drive, Docs, Sheets, Slides, Calendar, Chat, People) reject requests from any OAuth client that is not marked as agentic. The toggle is Google Cloud console → Google Auth Platform → Clients → ‘Use this client for an AI-powered agent’.
On November 12, 2026 the agentic setting becomes permanent on existing OAuth clients. If yours is not marked agentic by then, you cannot flip it later — you must create a brand-new OAuth Client ID, mark it agentic, and update every config that references the old client.
No. Apps and integrations that only call Google Workspace REST APIs — not the MCP servers — are not affected. The requirement applies only to OAuth clients calling the Workspace MCP servers.
Google’s guidance says no: maintain separate OAuth 2.0 Client IDs for human-interactive traffic (standard REST APIs) and AI agent workflows (MCP). Mark only the agent’s client as agentic. Tagging a client as agentic does not grant extra scopes or bypass user consent.
Google Cloud console → Google Auth Platform → Clients → select your OAuth client → check ‘Use this client for an AI-powered agent’. The setting has been editable since October 12, 2026.
AI Agent API Access Safety Check · AI Agents Negotiating Deals Check · OpenAI ExploitGym → Hugging Face Breach Check