Gmail’s most controversial new trick is not a flashy AI assistant or a redesigned inbox, it is a quiet default that lets the service scan your messages to power smarter tools and, crucially, to help train artificial intelligence. That automatic setting is switched on for you, not requested with a clear yes or no. If you care about who sees your private conversations, you should turn it off before you send another email.
I see this as the single most important privacy toggle in your account, because it sits at the intersection of convenience, data collection, and the fast‑moving AI race. Leaving it enabled means accepting that your inbox is raw material for machine learning, while disabling it is one of the few direct ways you can push back.
What the “smart features” AI setting actually does
At the heart of the current backlash is a cluster of “smart features” that let Gmail scan the content of your emails and attachments to generate suggestions, summaries, and other AI‑driven tricks. These tools sound harmless, but they rely on deep access to your inbox, and reporting shows that Gmail users are automatically opted in to allow Google to use that email data to train AI models. In other words, the same system that offers to finish your sentences is also learning from the full text of your conversations.
The key problem is not that AI exists inside Gmail, it is that Users are enrolled in this data sharing by default, with the burden placed on each person to find and disable the option. When I look at how this is structured, it is clear the setting is designed to feel like a simple preference about convenience, when in reality it is a consent switch for whether Nov Gmail can treat your inbox as a training set. That is the automatic setting I believe you should disable as soon as possible.
Why this default matters for 1.8 billion inboxes
Gmail is not a niche product, it is one of the core communication tools of the modern internet, and any change to its data practices instantly affects a huge slice of the planet. One detailed breakdown of the new AI behavior points out that we are talking about over 1.8 billion users, so any little tweak Google makes to Gmail is not just a UI experiment, it is a policy shift that touches almost every industry and community. When an AI training switch is turned on by default at that scale, it becomes a global privacy decision made on behalf of billions of people.
That scale is exactly why I see the default as so risky. A small business owner using Gmail to negotiate contracts, a doctor coordinating with patients, or a parent emailing about their child’s school all sit inside the same system that is now wired to harvest content for machine learning. When Nov Google quietly bakes AI training into the same settings that control your inbox layout, the stakes are far higher than a cosmetic redesign of Gmail.
Inside the setting that lets Gmail read your emails
To understand what you are opting out of, it helps to look closely at how the controls are described. Deep in the menus, there is a section labeled for Google Workspace smart features that spells out how Gmail can read your emails and attachments to power AI. The instructions explain that you need to be Still in Settings, locate the Workspace smart features area, Click on Manage Workspace smart feature settings, and then toggle both options off if you do not want your messages used in this way.
What jumps out at me is how many steps it takes to reach a simple “no.” You have to know that this category exists, drill into Settings, and then interpret language that blends productivity perks with data access. The structure makes it easy for a busy person to accept the default without realizing that they have effectively granted permission for ongoing scanning of their inbox. That is why I focus on this specific control as the one automatic setting you should disable first.
How to turn off Gmail’s AI training in a few clicks
Disabling the AI‑heavy smart features is not difficult once you know where to look, but the path is not obvious. The most direct route starts when you Open Gmail, click the gear icon in the top right, then choose “See all settings” to access the full configuration panel. From there, you go to the General tab and scroll until you find the smart features and personalization options that control how AI interacts with your messages.
Once you reach that section, you will see separate toggles for features inside Gmail and for related products like Maps or Wallet, which matches reporting that In January, Google updated its smart feature personalization settings so that you could turn off the features for Google Workspace apps and other services independently of each other. I recommend turning off every toggle that allows content from your inbox to be used for personalization or AI, then saving changes at the bottom of the page. It is a small investment of time that sharply limits how much of your email life is fed into training pipelines.
The “Important” label and features now tied to AI consent
One of the most troubling side effects of the new defaults is how core inbox functions are now entangled with AI permissions. Users have reported that Gmail is holding “Important” messages hostage behind new Smart Features AI permissions, describing how messages that should be flagged as Important no longer behave the way they used to unless those smart tools are enabled. In one widely shared account, a user explains that Since they discovered the change, they have had to choose between accurate importance markers and keeping AI scanning off, which feels less like a feature and more like pressure.
I see this as a subtle but powerful form of coercion. When a platform like Nov Gmail ties basic organization tools to the same setting that governs whether your emails can be mined for training, it nudges you toward accepting more data collection than you might otherwise allow. A detailed thread on Important messages makes clear that people are not imagining this shift, they are experiencing real friction when they try to keep AI off while still relying on the inbox behavior they have used for years.
How to find and disable related smart features across Gmail
The main AI training toggle is not the only place Gmail now leans on smart features, so I recommend doing a broader sweep of your account. One practical example comes from users who noticed a “happening soon” label tied to shopping emails, including Amazon pickup notices, that surfaced events in the inbox in ways they did not expect. The fix in that case is to go to Settings (gear) in GMAIL, choose All Settings, then go to or stay at the General Tab and look for Smart Features, where you can turn them off or set them to none.
If you have experimented heavily with Gmail over the years, you might also want a clean slate. One guide suggests a simple Method 1: Using the Settings Menu, which starts when you Open the Settings Menu from the gear icon After logging into your inbox and then walks you through restoring defaults. That step by step approach, outlined in detail by Method instructions, can help you undo older tweaks that interact strangely with the new AI options. I find that combining a reset with a careful review of every smart feature toggle gives you the best shot at an inbox that works for you without quietly feeding extra data into training systems.
Why Google’s AI race fuels aggressive data collection
To understand why these defaults are so assertive, you have to zoom out to the broader AI competition. The company behind Gmail is already facing intense scrutiny over its AI training practices, particularly as competitors like OpenAI and Microso race to build more powerful language models. In that context, every large, regularly updated dataset becomes a strategic asset, and few datasets are richer than the daily flow of personal and professional email.
That pressure helps explain why Nov Google is so eager to keep smart features turned on by default and why it frames them as essential to the modern Gmail experience. From my perspective, the company is trying to balance public assurances that it is not misusing data with product designs that still maximize the volume and variety of information available for training. The result is a system where the safest choice for your privacy, turning off AI‑driven scanning, is never the path of least resistance.
The privacy backlash and what critics are warning about
As more people discover how the new settings work, the backlash has grown sharper and more specific. One widely shared warning argues that Nov Google is harvesting 1.8 billion users’ private communications without explicit consent by making basic features dependent on AI scanning, effectively turning every inbox into a test pig for Google’s AI training. The same critique leans on the old line that if you do not pay, you are the product, arguing that the value you get in return does not justify the depth of access granted.
I find those concerns hard to dismiss. When critics describe Gmail making major changes in the way it used to work, they are not just talking about cosmetic tweaks, they are pointing to a shift in the social contract between Users and the platform. The automatic AI training setting is the clearest expression of that shift, and disabling it is one of the few concrete steps you can take to reassert control over how your email is used.
Sorting rumor from reality around Gmail’s AI training
The uproar around these settings has also produced confusion, with some viral posts overstating or mischaracterizing what exactly Gmail is doing. In response, Nov Google has publicly pushed back, arguing that some of the most alarming claims are a false alarm and that the company is not secretly expanding training beyond what its policies already allow. One detailed analysis notes that the company has tried to debunk misleading reports of Gmail using your emails to train AI in ways that go beyond the documented smart feature settings, even as it faces intense scrutiny over its broader AI practices.
From my vantage point, both sides of that argument matter. It is important not to attribute unverified behavior to Gmail, and I treat any claim that is not backed by the available reporting as “Unverified based on available sources.” At the same time, the confirmed facts are serious enough on their own: smart features do scan content, Users are automatically opted in, and the path to opt out is buried in Settings. That is more than enough reason to seek out the relevant controls and turn them off if you are not comfortable with your inbox doubling as AI training data.
More from MorningOverview