Google Rejects Claims That Gmail Emails Are Training Its AI

Google has denied reports circulating on social media and in tech outlets that it is using Gmail messages and attachments to train its Gemini AI models.

According to the company, no recent policy change has occurred, and existing privacy controls remain unchanged.

A Google spokesperson, Jenny Thomson, told The Verge:

“These reports are misleading – we have not changed anyone’s settings, Gmail Smart Features have existed for many years, and we do not use your Gmail content for training our Gemini AI model.”

The controversy began after a widely shared post on X by electronics YouTuber Dave Jones (EEVblog) claimed that users had been “automatically OPTED IN” to a setting allowing Gmail to read all private messages and attachments for AI training, unless manually disabled in two separate places. 

Malwarebytes followed with a blog post alleging that Google’s “Smart Features”, such as spell check, predictive text, and automatic order-tracking, are being used as a vector to collect personal content unless users explicitly opt out.

Google concedes that an update earlier this year allowed users to manage Smart Features separately across Google Workspace (which includes Gmail, Calendar, Docs) and other Google products like Maps and Wallet.

That change, the company says, may have contributed to the confusion, particularly after reports that some users discovered their previously disabled features had been re-enabled without their input.

Google describes Smart Features as tools that personalize a user’s Gmail experience, for instance, by enabling package tracking or adding flight details to calendars, but stresses that personalization does not equate to feeding email data into AI-training pipelines.

PolitiFact has also weighed in, noting that while Google’s Gemini Deep Research tool can access Gmail content if the user grants the app permission, that access is distinct from using email content to train the underlying model.

Still, some users report that the opt-out process is not straightforward.

So that's why we wrote step-by-step guides for you, because to stop AI training and fully opt out, you must disable Smart features in two different sections of your Gmail settings.

Location 1: Turn Off Main Smart Features

1. Open Settings: Go to your Gmail settings (gear icon → See all settings or Menu → Settings on mobile).
2. Find Section: Scroll to Smart Features in Gmail, Chat, and Meet.
3. Uncheck: Uncheck this option.
4. Save: Click Save changes (if on desktop).

Location 2: Turn Off Google Workspace Smart Features

1. Stay in Settings: Still in the settings menu, find Google Workspace smart features.
2. Manage: Click Manage Workspace smart feature settings.
3. Toggle Off: Turn OFF both options you see:
        Smart features in Google Workspace
        Smart features in other Google products
4. Save: Click Save.

Verify

• Confirm both sets of settings are off.
• Refresh Gmail or sign out/in to finalize.

If you follow these steps, you don't have to worry about whether Google is using your Gmail emails to train its AI models or not.

The issue has heightened scrutiny of Google’s AI-privacy practices just as the company is actively promoting Gemini 3, its latest advanced AI system and users are concerned about their privacy and data protection.