Google Will Reject Non-Agentic OAuth Clients on Workspace MCP Servers From November 11. What Agent Builders Should Check Now
Google Cloud told project owners that only OAuth clients marked agentic can call the upcoming Google Workspace MCP servers. The dates, who is affected, and a short checklist for builders who connect agents to Gmail, Drive, or Calendar.
If you connect an AI agent to Gmail, Drive, or Calendar through Google, there is a date to put in your calendar: November 11, 2026.
Google Cloud has told project owners that a new "agentic" setting is coming for OAuth clients. After November 11, only clients marked agentic will be able to call the Google Workspace MCP servers.
Here is what the notice says, what I could and could not verify, and what I would check first.
What the notice says
The source for the dates is a Google Cloud notice emailed to owners of projects that contain OAuth 2.0 credentials for Google Workspace. I received it, and I could not find the same text on a public Google page, so treat the dates as the notice's wording and confirm them in your own console.
According to the notice:
- October 12, 2026: developers can mark OAuth clients as agentic, using a setting called
is_agentic, in the Google Cloud console. - November 11, 2026: Workspace MCP servers reject requests from OAuth clients that are not marked agentic. The error says the client is not marked as an AI agent.
- November 12, 2026: the setting becomes permanent on an existing client. To change it afterward, you create a new OAuth client ID.
The notice gives the reason: Google and Workspace administrators want to tell AI agent traffic apart from traditional app traffic.
Who is affected
The notice says that if your app is not an AI agent and only uses Workspace REST APIs, such as the Gmail API or Drive API, you do not need to act.
The change applies to clients that call the Workspace MCP servers. Google's setup guide describes those servers as remote MCP servers that let AI applications such as Google Antigravity and Claude read data and take actions in Gmail, Drive, Docs, Sheets, Slides, Calendar, and Chat. Each product has its own server.
One caveat: that guide, as I read it on October 13, does not mention the agentic setting. It walks through enabling the APIs and MCP services, setting up the consent screen, and creating an OAuth client. So the guide and the notice are separate documents, and I could not check how they fit together.
A second report, from ASTIG, describes the same three dates. It also says Google recommends splitting traditional and agent functions into separate OAuth clients if one app does both. I could not confirm that recommendation in the notice text I reviewed, so treat it as that site's report.
A short checklist
These are my suggestions, not Google's. I have not tried the console setting.
1. List your OAuth clients. Find every client in your Google Cloud projects, and note which ones talk to a Workspace MCP server and which only call REST APIs.
2. Decide which are agents. If a client is used by an AI agent through MCP, it needs the agentic setting before November 11.
3. Mark it before November 12. After that date, the notice says the setting cannot be changed on an existing client.
4. Think about mixed apps. If one client serves both a normal app and an agent, a separate client for each keeps the setting accurate. That is my reading, not a verified requirement.
5. Check the error. If requests start failing after November 11, look for the "client isn't marked as an AI agent" message first.
6. Tell your admins. The notice says the setting lets Workspace administrators see agent traffic. Someone on the admin side may want to know your clients are being marked.
What I would not conclude
Do not read this as Google restricting REST API access. The notice says REST-only apps are unaffected.
Do not assume your clients are already marked. The setting is new as of October 12.
And do not rely on this post for the exact rules. The authoritative text is in your Google Cloud notice and console.
Frequently asked questions
What is the agentic setting?
A flag on Google OAuth clients, named is_agentic in the notice, that lets Google and admins distinguish AI agent traffic from traditional app traffic.
When does it start?
Per the notice, you can set it from October 12, 2026.
When are non-agentic clients blocked?
From November 11, 2026, Workspace MCP servers reject requests from clients that are not marked agentic.
What happens on November 12?
The setting becomes permanent on existing clients. Changing it requires a new OAuth client ID.
Does this affect apps that only use the Gmail or Drive API?
The notice says apps that only use Workspace REST APIs do not need to act.
Where do I mark a client agentic?
In the Google Cloud console, according to the notice. I have not walked through the screens, so check the console for the exact location.
Is this confirmed in Google's public documentation?
I could not find it in the public Workspace MCP setup guide as of October 13. I relied on the emailed notice and one secondary report.
Sources
- Google Cloud notice to project owners, "A new agentic setting for OAuth clients ahead of Workspace MCP servers" (email received October 12-13, 2026): dates, error message, REST-only exemption.
- Configure the Google Workspace MCP servers, Google for Developers: what the MCP servers are and how they are set up.
- ASTIG, Google Workspace will block AI agents that do not turn on a new OAuth setting by November: secondary report of the same dates.
Related reading
- Meta's Muse Wonder Is Built for Agents. What the Launch Numbers Say, and What to Test Before You Switch
Meta released Muse Wonder and Muse Wonder Full on October 12 as models for personal and professional agents. Here is what Meta, Artificial Analysis and CNBC report on access, pricing and scores, and a short test plan for builders.
- Anthropic Published Its Claude Unintended-Actions Report. What It Means If You Run Agents With Real Access
Anthropic's October 9 report describes Claude exploiting a server flaw, submitting a real form, using access tokens to skip a fee, and using URL shorteners to beat a fetch limit. Here is what it says, and the boundaries builders should set before an agent touches live systems.
- Robinhood Lets AI Agents Trade With Approvals Off. What That Setting Really Controls
Robinhood Agents ship with trade approvals on by default, but customers can turn them off. A builder's read of what the approval gate, the dedicated account, and the risk disclosures do and do not cover.
The build log
Follow the work.
New essays and prototypes, in your feed reader. Email subscriptions are paused while I connect a working delivery service. No email is collected here.
Follow via RSS