Disable Gemini

Google Gemini is a so-called "Artificial Intelligence" or "AI" service offered by Google. It integrates with several Google Workspace services and is enabled by default. As the Google Workspace administrator you can disable it using the instructions here. (Correct as at 13 February 2025)

Note that the instructions for you will depend on which edition of Google Workspace you are using because Google has not (yet) enabled the ability to control access to Gemini in all editions without intervention by Google Support.

To get started, you'll need to log-in as the Google Workspace Administrator.

Navigate to "Generative AI" -> "Gemini for Workspace"

Once you do, you'll see one of these images.

Gemini Workspace Settings unavailable

If you do not have access to Gemini Workspace Settings in your Workspace panel, you need to contact Google Support to enable it.

Click on the help (?) icon at the top right.
Ask the support bot to "Disable Google Gemini".
It will offer meaningless advice.
Keep clicking "This doesn't help".
Eventually it will open a chat session with a Google Support person.
Tell the person that you cannot disable Google Gemini on your workspace because there are no controls.
Provide a screenshot.
Wait until they enable the controls.
Provide feedback that this should be opt-in, not opt-out and that these controls should be enabled for every single workspace account.
Google Support will potentially advocate for the use of Gemini, feel free to engage in debate at will.

Gemini Workspace Settings enabled for all services (the default)

Once you can see the Gemini Workspace Settings, you can edit them for each individual service.

It's likely that this list will grow and I suspect that it will enable Gemini for each service by default, so you might need to do this more than once.

Click on the pencil icon and edit each individual service.

At time of writing these services will need editing individually (*):

Drive and Docs
Gmail
Google Chat
Google Meet

(*) Gemini App is not listed above, it's controlled on a different screen entirely, see below.

Gemini Workspace Settings disabled for all services

After you've disabled Gemini for all your services, they should show as "off".

You can click the pencil icon to enable them if you want to.

Gemini App

Note that these instructions are NOT related to the Gemini App, which must controlled on a separate screen, in the same "Generative AI" section, under "Gemini App".

This setting controls your users access to the gemini.google.com application.

It should be available to all Google Workspace Editions.

Why does this matter?

Without going into phylosophical arguments about the nature of "AI", there are several reasons why you should consider the reprecussions of Google Gemini being enabled on your Google Workspace account.

This article published on 12 February 2025 (https://arstechnica.com/security/2025/02/new-hack-uses-prompt-injection-to-corrupt-geminis-long-term-memory/) shows the first published exploit of Google Gemini and sounds the warning bells of things to come. Essentially, there is no difference between "information" and "instruction" in the current "AI" offerings. This means that a malicious email sent by anyone could, at least theoretically, include instructions intended specifically for Google Gemini. These so-called "injections" have been increasing over the past months and there is, in my opinion, no way to mitigate against this class of attack. The only option available at this point is to fix each individual attack.
Google has unilaterally enabled this service across all your Google Workspace data. As an organisation you have had no opportunity to evaluate the impact of this and the implications of all your data being included in any "AI" process. Whilst Google already has all your data, it's an entirely different thing for your data to be used across various services. Consider for example a confidential spreadsheet containing information that is pertinant (by "AI" standards) to an email you are drafting to someone who should not have access to this data. How do you as an organisation protect against this if you're unaware that this is happening at all. Whilst you can argue about the merits of "AI", making this an opt-out is not appropriate for a corporate environment.

Report abuse