Google is pushing Gemini into Gmail, Drive, and the rest of Workspace while routing the privacy opt-outs through menus that punish users for using them. The company expects to spend $185 billion on AI in 2026 alone, and the user-facing math of that investment shows up in defaults: Gemini is on, the toggle to turn it off is hard to find, and flipping it disables unrelated features that have nothing to do with generative AI.
The clearest case is in Gmail. To disable Gemini, a user has to turn off what Google calls Smart Features. There is no Gemini-specific switch. The Smart Features toggle also controls inbox tabs, Smart Compose, package tracking, and Calendar event extraction from email — features that predate the AI boom by years.
The cost is immediate. A Gmail account that showed 20 unread messages with social and promotional tabs filtered can show 500 unread the moment Smart Features goes off, because every newsletter and shipping update is suddenly back in the primary inbox. After the toggle flips, Gmail surfaces a pop-up offering to switch everything back on, Gemini included.
Key facts
- 01Google expects $185 billion in AI-related spending in 2026 alone, the backdrop for Gemini's expansion across Gmail, Drive, and Workspace.
- 02Disabling Gemini in Gmail via Smart Features also kills inbox tabs, Smart Compose, package tracking, and Calendar event extraction.
- 03Turning off the Smart Features toggle can flip a user's inbox from 20 unread to 500 unread as filtered social and promo mail floods the primary tab.
- 04The Gemini Apps Activity opt-out also wipes chat history, forcing users to pick between training data and product utility.
- 05Dr. Harry Brignull, who coined 'dark pattern' in 2010, calls Google's approach a textbook pre-selection pattern.
There are two Smart Features toggles, one in Gmail settings and a second a click deeper for Workspace. The Workspace version is supposed to remove Gemini from Drive, but the Gemini UI elements remain visible after the switch is off. Clicking any of them produces a prompt asking the user to re-enable Smart Features.
“Google expects to spend $185 billion on AI in 2026 alone, and the path to opting Gemini out of your Gmail runs through menus that quietly disable inbox tabs, Smart Compose, and package tracking.”— Jaeden Schafer
Opting out of training data is its own maze. Google says in a recent blog post that it does not train its foundational models on the contents of Gmail or Drive, and a Google spokesperson said that 'protecting users' privacy and control over their data is fundamental to how we develop and deploy AI in Google Workspace.' But Gemini's own inputs and outputs — which can include summaries and snippets of user emails and files — are eligible for training unless the user disables Gemini Apps Activity.
Disabling Gemini Apps Activity also deletes the user's Gemini chat history and stops new chats from being saved. The choice on offer is binary: keep a usable assistant with memory, or keep training data out of Google's pipeline. There is no middle setting.
Finding the toggle is harder than it should be. The Gemini privacy menu is not linked from Google's account privacy settings where users would expect it. A Google representative told Ars Technica the link should appear inside Activity Controls alongside submenus for Android, Maps, Search, Assistant, and Wallet. Ars checked multiple accounts and could not find it.
Marie Potel of Fair Patterns, a startup building models that detect predatory UI design, said intent is beside the point. 'It doesn't matter whether it's intentional or not,' Potel said. 'What matters is whether the autonomy — the agency — of users is respected and whether the design goes against what users want to do.' Potel added that 'Google has a history of hiding features, especially privacy settings, in a number of clicks that are absolutely made to deter people from using them.'
Dr. Harry Brignull, who coined the term 'dark pattern' in 2010, framed the broader playbook as pre-selection. 'If you want to get people to opt in to something, instead of asking them in a sign-up step explicitly with a link to learn more, you can take that and make it pre-selected,' Brignull said. Defaults compound: most users never touch them, and Google knows it.
The counter-argument from Google is that the company has been transparent about what Gemini does and does not see, that personal Drive content is not used to train foundational models, and that users retain controls. That is technically accurate. It is also compatible with a flow in which the controls are buried, the opt-out costs unrelated features, and the model still trains on Gemini outputs that quote the user's own data back to them.
The Gemini rollout sits alongside Google's other 2026 AI moves — Google Cloud cleared $20 billion in Q1 and said it could not keep up with demand, and Google added 25 million paid subscriptions in the same quarter. Gemini engagement is the through-line that justifies the capex line. The more Gmail and Drive users who never turn the assistant off, the easier the $185 billion number is to defend to investors.
That is the real shape of the privacy story. Gemini's defaults are not an accident of busy product managers; they are the user-acquisition strategy for a model business that needs scale to pencil. Until disabling AI features in Gmail stops costing users their inbox tabs, the choice Google offers is not really a choice — it is a tax on opting out, paid in friction.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
