Does Google AI read my Gmail? What Gemini sees and opt-outs

Every few months a headline says Google's AI is reading your inbox, then Google says it isn't, and both are partly right. They are talking about different things. Here is what actually happens to your Gmail, according to Google's own documentation, and how to turn each piece off.

Short answer: yes, Google's AI processes the contents of your Gmail. Smart features sort your mail and write suggested replies, and Gemini summarizes threads and answers questions about your inbox. Google says it does not train Gemini directly on your inbox. But if you connect Gmail to the Gemini app, Google's help pages (checked October 16, 2026) say summaries and excerpts of your emails can be used to train its models and can be seen by human reviewers. All of it is controlled by settings, and many accounts have them switched on.

Three ways Google's AI touches your Gmail

"Google's AI" is not one thing. There are three separate systems, each with its own setting, and most guides mix them up.

The first two come straight from Google's smart features help page, which lists these examples and the regions where the settings start switched off. The third launched on January 14, 2026 as a beta for paying Google AI Pro and Ultra subscribers in the US, and Google said it is "off by default and you can choose specific apps to connect," as 9to5Google reported. Google's eligibility page now says it is "being released gradually," needs a personal account, age 18 or over and Keep Activity switched on, and is not available in the EEA, Nigeria, Switzerland or the UK.

The AI inside Gmail itself has grown fast. On January 8, 2026 Google announced Gmail was "entering the Gemini era," with AI Overview thread summaries, Help Me Write and suggested replies rolling out free to everyone, starting in US English, per Google's announcement. The full Gemini assistant in Gmail, which can search and organize your inbox, requires an eligible Workspace or Google AI plan.

Does Google train Gemini on your emails?

Google says not directly. A Google spokesperson, Jenny Thomson, told The Verge in November 2025: "we do not use your Gmail content for training our Gemini AI model." Gemini inside Gmail follows that rule. The fine print matters in two places.

First, "not training Gemini" is not the same as "not used at all." Google's smart features page says that if you have turned on any of the smart features settings, "we may also process your Workspace Content & Activity to improve these features." It also says "the learnings developed from this improvement process may persist" after you switch the setting off.

Second, the Gemini app plays by different rules once you connect Gmail to it. Google's page About personalization with Connected Apps says "Gemini does not train generative AI models directly on your Gmail inbox". The next sentences describe what it does train on:

• "summaries, excerpts, generated media, and inferences from your relevant media, emails, files, and transactions"
• and, in its own words, "In some cases, for example, where the email or file is brief or particularly relevant to respond to the prompt, the summary may just be the email or file."

Your prompts and Gemini's answers are training data too, and Google's privacy policy, in the version effective October 12, 2026, says it uses "your interactions with AI models and technologies like Gemini Apps to develop, train, fine-tune, and improve these models." So the honest version is this: your inbox is not the training set, but a short email Gemini quotes back to you can end up in it. Google's own advice on that page is blunt: "Please don't connect apps to Gemini if they contain personal or confidential information that you wouldn't want to be used for training generative AI models."

Can a human at Google read your Gmail?

Not as part of Gmail's own features. They run automatically, and Google's Gmail privacy page says: "We access your private content only when we have your permission or are required to by law."

The Gemini app is the exception you opt into. Google says "a subset of Connected Apps data processed by Gemini is reviewed by human reviewers," including "summaries, excerpts and inferences from your relevant emails and files." Google says it first disconnects these from your account and reduces personal information in them, and that if you send feedback, "human reviewers may also review some of your Gmail inbox." Reviewed chats are kept for up to three years, per the Gemini Apps Privacy Hub.

That is a real set of limits. It is also policy, and the mail sits on Google's servers in a form Google's systems can process, which is why every one of these features can exist in the first place. We cover the rest of that picture, from legal requests to third-party apps, in is Gmail private?

How to turn off Google's AI in Gmail

Full opt-out takes three steps, because there are three systems. Each is a different switch, so do all three.

On a computer

1. Smart features in Gmail, Chat, and Meet. In Gmail on the web, click the gear, then See all settings. On the General tab, find Smart features and untick "Turn on smart features in Gmail, Chat, and Meet." Click Save changes at the bottom.
2. Workspace smart features. On the same tab, click Manage Workspace smart feature settings and switch off both "Smart features in Google Workspace" and "Smart features in other Google products." Save. The first covers Gemini summaries and drafts inside Gmail. The second is the one the Gemini app needs: Google says "Gemini Apps can't connect to Google Workspace when this setting is off."
3. The Gemini app. Go to gemini.google.com, click Customize in the sidebar (under More if you don't see it), then Connectors at the top. Find Google Workspace, click More options, then Disconnect. If you also turn off Keep Activity, Google says Workspace apps become unavailable to Gemini and your future chats are not used to improve its models unless you send feedback.

On Android: open Gmail, tap Menu, then Settings, and pick your account. Under General, untick Smart features, then tap Google Workspace smart features and untick both boxes.

On iPhone or iPad: open Gmail, tap Menu, then Settings, then Data privacy. Turn off Smart features, then tap Google Workspace smart features, turn off both, and tap Done.

These menu names are from Google's help pages as of October 16, 2026. Google notes that some changes made on a computer don't show in the mobile app, so check both.

To check it worked, reload Gmail. The Primary/Social/Promotions tabs disappear and Smart Compose stops offering text. That is the price: you lose the tabs, the summary cards, the suggested replies and the AI summaries. Google built it as a trade, and you are the one who decides.

One caution from the last time this blew up. In November 2025, Malwarebytes published a story saying Gmail was reading emails to train its AI, then corrected it, saying Google had reworded and moved settings that were not actually new. The settings are real and worth reviewing. The "secretly training on your inbox" version was not supported.

The Gemini lawsuit: Thele v. Google (status as of October 2026)

The fight over whether Google switched these features on without asking is now in federal court.

• The claim. Thomas Thele, later joined by Melo Porter, sued Google in the Northern District of California on November 11, 2025 (case 5:25-cv-09704). They allege that around October 10, 2025, Google turned Gemini on by default across Gmail, Chat and Meet, letting it reach private messages and attachments without consent, in violation of California's Invasion of Privacy Act, the Stored Communications Act and other laws.
• The dismissal. On July 7, 2026, Judge Noël Wise granted Google's motion to dismiss, but with leave to amend. She wrote that "plaintiffs have so far only alleged that Gemini could be used to track their data. This is insufficient to allege an injury in fact," as Bloomberg Law reported. In plain terms, the court did not find that Google behaved well. It found the plaintiffs had not shown Gemini actually touched their data.
• Where it stands now. The plaintiffs, now joined by a third, Edward Goldstein, filed a second amended complaint on August 18, 2026. Google moved to dismiss it again on September 29, 2026, with a hearing set for January 13, 2027. A case management conference that had been set for October 13, 2026 was taken off calendar on September 30 and reset to March 2, 2027, according to the court docket. As of October 16, 2026, there is no ruling on the new complaint, no certified class and no settlement. If someone offers you a "Gmail AI settlement" claim form, it is not from this case.

We will update this section as the case moves.

What the case already shows is where the argument sits. Nobody disputes that Gmail can process your messages. The dispute is whether a switch was flipped for particular people, and whether they can prove it. As long as your provider can read your mail, the question of who gets to read it will keep coming back as a settings change, a policy update or a lawsuit.

If you want an inbox that is not mined for AI

Switching the features off limits what Google does with your mail today. It doesn't change the fact that Google can process it, and a setting can be redesigned or moved.

The alternative is email built so the provider has nothing to mine. That is what we built Secria Mail to be. Your keys are created on your device and only reach our servers encrypted, and stored mail is under zero-access encryption. There is no AI toggle to hunt for, because Secria never mines your mail. Every plan, including the free one, uses hybrid ML-KEM-1024 and X25519 post-quantum encryption. If you are weighing options, our guide to the most private email service sets out the criteria we would use.

FAQ

Is Google reading my emails to train its AI?

Not directly, according to Google. It does process your mail to run smart features and may use that to improve them. If you connect Gmail to the Gemini app, Google says it trains on summaries and excerpts of relevant emails, and sometimes the summary is the email itself.

Can Gemini search my Gmail?

Yes, in two places. The Gemini assistant inside Gmail can search your inbox on an eligible paid plan, and the Gemini app can search it if you connect Google Workspace through Personal Intelligence. Turning off "Smart features in other Google products" in Gmail blocks the second one.

Does Google read your Gemini chats?

Some of them. Google says a subset of chats are reviewed by human reviewers, including trained service providers, after being disconnected from your account. With Keep Activity on, chats are also used to train its models. Reviewed chats are kept for up to three years even if you delete your activity.

Does turning off smart features stop Gemini in Gmail?

Yes, if you turn off both Workspace settings. "Smart features in Google Workspace" covers Gemini summaries and drafting inside Gmail, and "Smart features in other Google products" stops the Gemini app from connecting to Gmail. Disconnect Google Workspace in the Gemini app as well, so nothing reconnects later.

Why are smart features off in my account already?

Google turns these settings off by default for people in the European Economic Area, the UK, Switzerland and Japan. If you are elsewhere and never saw a prompt, they may be on, so check the General tab in Gmail settings.

Did Google win the Gemini Gmail lawsuit?

It won the first round. The judge dismissed the case in July 2026 because the plaintiffs had not shown their own data was accessed, and allowed them to try again. They filed an amended complaint in August 2026, Google has asked the court to dismiss it again, and a hearing on that is set for January 2027, so the case is still open.

Secria Mail keeps every message in your mailbox under zero-access encryption and never mines your mail. The free plan needs no card.