If you use an Android phone, there’s a good chance Google’s Gemini AI is now interacting with your apps, even if you thought you had disabled it. The company recently rolled out changes that grant Gemini new levels of access to messages, phone calls, and third-party apps like WhatsApp, regardless of whether users had previously opted out. If that sounds invasive, it’s because it is.
In emails sent to Android users ahead of the rollout, Google framed the change as a convenience upgrade. Gemini, they said, can now help users perform everyday tasks more easily, such as initiating calls, sending messages, and launching utilities. These functions, Google explained, would be available “whether your Gemini Apps Activity is on or off.” The update began rolling out automatically on July 7.
This line, buried halfway through the notice, immediately clashed with another sentence just a few paragraphs later: “If you have already turned these features off, they will remain off.” Which is it? Is Gemini respecting user settings or overriding them?
Nowhere in the announcement, or in any of the linked support documentation, did Google clearly explain how to remove Gemini entirely or prevent it from gaining access to sensitive apps. Some users were left combing through settings. Others, depending on device brand, location, or operating system version, couldn’t find the Gemini app at all.
Gemini, which is Google’s next-generation AI assistant and the eventual replacement for Google Assistant, already had access to user data in various ways through “extensions.” These extensions allowed Gemini to interact with services like Google Keep, Gmail, and YouTube to provide customized responses. The problem, critics say, is that this customization comes at the cost of privacy. And it’s not always clear where the line is drawn.
According to official documentation, Gemini can access system permissions like call logs, contacts, message history, and screen content to provide more “context-aware help.” That might be tolerable for some users who have opted in. But under the new changes, that line becomes murky.
Even users who haven’t interacted with Gemini, or who assumed it was inactive, are finding that it’s more deeply embedded than expected. One user on a Pixel 7 device reported trying to access Gemini settings only to find the app seemingly absent. Yet according to Google’s own policy, Gemini can operate without saving any visible activity beyond a 72-hour buffer. That raises the question of whether “off” really means off.
Another concern is how data is handled once Gemini has access. The company admits that “human reviewers (including service providers)” can read and annotate user activity, including chats, files, and voice recordings. All of it is fair game if Gemini is active. That includes interactions that happen when Gemini extensions are used, when the AI is invoked from within an app, or when it’s running quietly in the background.
When one user contacted Google directly for instructions on fully disabling Gemini, the company responded by repeating marketing language about how helpful the AI could be. They offered the same vague links that were already causing confusion. There was no clear answer about how to ensure that Gemini integrations are actually disabled or whether they can be at all.
The only relatively effective workaround came not from Google but from Tuta, a privacy-focused email and calendar provider. In a blog post, they explained that users may be able to block some data collection by disabling Gemini app activity and ensuring that the Gemini app isn’t installed on their device. But even that is incomplete. Gemini doesn’t always appear as a standalone app. On some phones, it may arrive baked into system updates.
Advanced users may try to uninstall Gemini manually using the Android Debug Bridge, a command-line tool typically used by developers. But this process is far from user-friendly. It requires enabling developer mode, installing software on a computer, and entering terminal commands. Even then, success isn’t guaranteed. In at least one case, an attempt to uninstall the Gemini package resulted in a vague system error with no indication of whether the app had ever been installed in the first place.
The lack of transparency is what makes this situation different. This isn’t just a case of an app being pushy or permissions being hard to manage. This is a major tech company quietly expanding the footprint of its AI engine deep into user devices, without offering a clear way out. For many Android users, especially those outside the United States or using phones from non-Google manufacturers, the experience is even more confusing.
Some have drawn comparisons to Microsoft’s bundling of Internet Explorer with Windows in the late 1990s. That move led to years of antitrust litigation. The core issue then was dominance through forced integration. Google may be heading down a similar path with Gemini, positioning it as a must-use assistant regardless of user preference.
It’s true that some people will find these features helpful. Hands-free message dictation, smart reminders, and real-time suggestions can make life easier. But the problem isn’t what Gemini can do. The problem is how quietly it’s doing it, and how difficult Google makes it for people to say no.
With AI rapidly becoming central to how tech companies design software, the balance between convenience and control is under pressure. Gemini is just one example of how companies are blurring that line. Unless users push back or regulators step in, it may become increasingly difficult to keep that control at all.
For now, if you’re concerned about Gemini’s presence on your phone, your best bet is to open your Google settings, dig through the privacy controls, and disable every toggle you can find. If the app isn’t installed, you might be safe. But there’s no guarantee. And unless Google starts offering a straight answer, you’ll probably have to keep checking.
