Google Just Turned On AI Access to Your Workspace — By Default

Holly Hartman
Apr 22
7 min read

SHADOW AI IN YOUR TECH STACK · SPECIAL ALERT

What the Workspace Intelligence rollout means for your organization, why “default on” is a governance event, and what you should consider doing right now.

By Holly Hartman, AIGP · FWS Enterprise LLC · April 22, 2026

This morning, Google Workspace administrators received an email. The subject was straightforward. The content was polite. And inside a few paragraphs of product language was a governance decision that affects every person in your organization effective today.

Google has launched Workspace Intelligence, an underlying AI system that gives Gemini real-time, cross-platform access to your organization’s Gmail, Drive, Chat, and Calendar. And unless an administrator actively changes it, it is on by default for every user in your organization starting April 22, 2026.

Before we go further, something important needs to be said clearly: Google is being transparent here. They sent the email. They’re providing admin controls. They’re explaining what is and isn’t changing about data handling. That matters, and it deserves acknowledgment. Transparency from a vendor is the right move and it’s exactly what responsible AI deployment should look like.

But transparency from the vendor doesn’t replace governance from you. Let’s talk about what this actually means.

Example Email sent April 22. 2026

What Workspace Intelligence Actually Does

In plain language: Workspace Intelligence is an AI layer that grounds Gemini’s responses in data from your organization’s Gmail, Drive, Chat, and Calendar. Instead of a user having to paste in context manually, Gemini now draws on those sources automatically — for the sources that are enabled.

Google’s own example is helpful: ask Gemini in Docs to draft a project update, and it will pull your meeting notes, relevant files, and chat threads about that project — then generate the report with citations.

That sounds useful. In many ways it is. But from a governance lens, what you just described is an AI system with broad, active access to your organization’s most sensitive communication channels — running by default, without employees needing to take any action to enable it.

“Default on” is not a product feature. It is a governance posture. And every organization gets to decide if it’s the right one for them.

Let’s Talk About “Default On”

This is the part that deserves your full attention as a leader or administrator.

In AI governance, default settings carry enormous weight. Research consistently shows that the vast majority of users — and organizations — never change defaults. When something is on by default, it is, for most practical purposes, always on.

What “default on” means here:

• Every user in your organization, from day one, has Gemini actively indexing their email, files, chats, and calendar

• No employee action was required to enable this

• Employees may not know it changed at all — Google’s notification went to administrators, not end users

• The Admin console controls to manage Workspace Intelligence may take up to 3 days to become visible after rollout begins — the feature itself may already be active

⚠ GOVERNANCE GAP WINDOW

Google has noted that the new Admin console controls may take up to 3 days to become visible after rollout begins. That means the feature may already be active in your organization before you have a way to adjust its settings. This is not a crisis — but it is a reason to act quickly and document your intent.

The Governance Flags You Need to See

01 — Employee Awareness Gap

Google’s notification went to administrators, not end users. Most employees have likely not been told that Workspace Intelligence is now grounding Gemini’s responses in their email, files, chats, and calendar. Whether you consider that a gap is a governance judgment call but awareness and consent are foundational governance principles.

02 — Contextual Integrity vs. Access Permissions

Google confirms that Gemini only surfaces content a user already has permission to view. That sounds reassuring but permissions and appropriate context are not the same thing. An employee might have access to a sensitive HR thread without expecting it to be pulled into an AI-generated draft.

03 — Third-Party Connected Sources

The email mentions Workspace Intelligence also draws from “connected third-party sources.” That scope is undefined in the announcement. Before any other action, administrators should audit what third-party applications are connected to their Workspace environment.

04 — Soft Walls, Not Hard Walls

If you disable a data source — say, Drive — users can still manually invoke it in a prompt. Gemini will use it. Turning a source “off” stops active indexing but does not prevent direct user-invoked access. Your governance approach needs to account for this distinction.

05 — AI Policy Alignment

Does your current AI use policy address employer-operated AI systems that access employee communications? Many organizations have policies focused on what employees do with AI — but fewer have addressed what the tools themselves are doing behind the scenes.

What Google Said That Deserves Credit

Good governance requires intellectual honesty — including when vendors do things right. Google’s communication included several commitments worth noting:

• Data is not used to train AI models, explicitly stated

• Data is not used for advertising purposes, explicitly stated

• Granular admin controls are being provided at the organizational unit and group level

• User-level content permissions are respected, Gemini won’t surface content a user doesn’t have access to

• The email was sent proactively, before the rollout took effect

This is what transparent AI deployment looks like. The governance obligation that follows is yours not because Google did something wrong, but because your organization’s AI posture is always your responsibility, regardless of what the vendor does or doesn’t do.

THE FWS PERSPECTIVE

This is a textbook Shadow AI event — not because Google is hiding anything, but because this AI capability now lives inside your organization’s most sensitive workflows without employees realizing it changed. The system was already in your tech stack. The default just changed.

This is exactly the kind of infrastructure shift that a mature AI governance posture is built to catch, evaluate, and respond to — not panic over, but not ignore either.

Actions to Consider Right Now

Immediate (This Week)

1. Check your Admin console for the new Workspace Intelligence controls. If they’re not visible yet, set a reminder to check again in 24–48 hours.

2. Audit your connected third-party sources. Go to Admin console → Apps → Google Workspace Marketplace apps and review what is connected.

3. Decide your default posture. Do you want this on org-wide? Restricted by department or user group? Off until your policy catches up? Make that decision deliberately, not by inaction.

4. Notify HR and Legal that this change occurred today. They should know before an employee brings it to them.

Short-Term (Next 30 Days)

1. Review your AI use policy for gaps around employer-operated AI systems. Most policies were written for what employees do — not what the tools do.

2. Draft an internal communication to employees explaining what Workspace Intelligence is, what it can access, and what controls exist. Transparency inside your organization matters as much as it does from your vendors.

3. Document your governance decision — whatever it is. Whether you leave it on, restrict it, or turn it off, write down why. That documentation matters if you ever need to demonstrate governance due diligence.

4. Add this to your AI Governance Committee agenda — or, if you don’t have one yet, let this be the moment that makes the case for one.

The Bigger Picture

What happened today with Google Workspace is going to keep happening. Microsoft Copilot. Salesforce Einstein. HubSpot Breeze. The platforms your organization already relies on are adding AI layers, often enabled by default, often without a direct notification to the people whose work is being processed.

This is not a reason to be afraid of these tools. Many of them are genuinely useful and thoughtfully built. It is a reason to have a governance infrastructure that can evaluate each one clearly, make deliberate decisions, and communicate those decisions to your people.

Being AI-ready doesn’t mean saying yes to everything. It doesn’t mean saying no to everything either. It means having the framework, the awareness, and the governance posture to make the call — on purpose, not by default.

“Your organization’s AI posture is determined by the decisions you make — and the ones you let someone else make for you.”

NOT SURE WHERE YOUR ORGANIZATION STANDS?

Take the FWS AI Readiness Quiz

Events like this one reveal exactly where governance gaps exist. The FWS AI Readiness Quiz helps you identify where your organization is on the journey from AI-anxious to AI-ready — and what your most important next steps are.

futureworkforcesystems.com/ai-readiness-quiz

Disclosure: This post was written by Holly Hartman, founder of FWS Enterprise LLC. FWS Enterprise is an AI governance and workforce readiness consultancy. This post reflects Holly’s professional analysis and perspective as an AI governance practitioner. FWS Enterprise has no financial relationship with Google or Google Workspace. Information is based on the Google Workspace administrator communication issued April 22, 2026, and is accurate to the best of the author’s knowledge at time of publication. Organizations should consult their own legal and technology counsel for guidance specific to their situation.

Frequently Asked Questions

Is Google Workspace Intelligence on by default?

Yes. As of April 22, 2026, Workspace Intelligence is enabled by default for all Google Workspace organizations. Administrators do not need to take any action to activate it — but they can restrict or disable specific data sources through the Admin console.

Can administrators turn off Workspace Intelligence?

Administrators can disable specific data sources — Gmail, Drive, Chat, or Calendar — at the organizational unit or group level. Disabling a source stops Gemini from actively grounding responses in that source. However, users can still manually invoke a disabled source in a prompt, so disabling creates a soft restriction, not a hard block. New Admin console controls may take up to 3 days to become visible after rollout begins.

What data does Gemini access through Workspace Intelligence?

By default, Workspace Intelligence grounds Gemini’s responses in content from Gmail, Google Drive, Google Chat, and Google Calendar, as well as connected third-party sources that vary by organization. Gemini only surfaces content the individual user already has permission to view — it does not cross user permission boundaries.

Does Google use Workspace Intelligence data to train AI models?

No. Google has explicitly stated that data processed through Workspace Intelligence is not used to train generative AI models and is not used for advertising purposes. Organizations should review Google’s current data processing terms and consult legal counsel for compliance requirements specific to their situation.

Why does this matter from a governance perspective?

Default-on AI systems that access sensitive communications represent a governance event regardless of vendor intent. Most employees were not directly notified of this change. Organizations without a current AI use policy, governance committee, or data source audit process are now operating with an active AI layer inside their most sensitive workflows — and may not have the infrastructure to evaluate, communicate, or manage it. That gap is the governance concern.