Gemini permissions to access files on device

hello, can we add the option to allow Gemini the optional permissions to access your files/devices?

That would be challenging as Gemini lives within your browser, and most browsers put the sites within a sandbox with a lot of limitations. It would be a serious security risk if any site can access your files directly from the browser.

Instead, you can try and allow Gemini access to your Google Drive or other online drive systems. (Dropbox, OneDrive, etc.) But even that could be a security risk. It would make things easier, but it comes with unacceptable risks…

For AI Studio, I have considered making my own API to access files in a specific folder for the apps to use. Such a system would work nicely, and I have already given my app read-only access to a folder with my photo collection. Now it can browse through all those photos without problems. But if I provide write-access to the same folder then a hacker could easily delete all my photos. And that’s a risk I won’t take. :slight_smile:

1 Like

Hello,

Thanks for the reply! I appreciate your concerns about this situation.

I can see that this might be challenging for you but, I assure you the security were working on and developing are sure to meet these concerns.

It shall be optional, for all to choose thier choice. If they don’t like it, they don’t like it. However, I would get much use out of it and be able to do much more very effective and effecintly.

I would enjoy to be able to tell my phone what to do and it go thru the motions speeding up productivity time.

1 Like

Subject: Understanding the Dev-Tool vs. Production Environment

"I think there’s a fundamental misunderstanding here regarding the role of AI Studio. It is a prototyping workbench, not a final application deployment environment.

If you’re looking for a consumer-facing AI that reads your local downloads folder, you’re looking for the Gemini App/Workspace integrations. However, as developers, we should be thinking about Production Architectures:

AI Studio is for Iteration: It’s where you test if the model can handle your specific data structures.

Firebase/GitHub is for Deployment: For a ‘Final Build’ or a personal project with persistent file access, you wouldn’t rely on a manual file upload button. You’d use Firebase Cloud Storage or a GitHub URI to feed data into the Gemini API programmatically.

The Solution: Instead of waiting for a ‘permission fix’ that likely won’t come to a browser-based IDE for security reasons, just host your associated files in your own Firebase instance or a private Repo. This gives your project a persistent, cloud-based memory that works across any device, rather than being tethered to a single local machine.

Let’s stop treating the ‘Workbench’ like the ‘Final Product.’ Use the right tool for the mission phase."

~CMD. Proton :vulcan_salute:

1 Like

Ok..

I believe I understand that aistudio is to develop and another to deploy?

I didn’t mention anything about that.

I mentioned the fact that Gemini should be able to access files on device/account, optionally.

I understand there will need to be integration between applications to make an application that’s deployed with monetization.

I have 50+ badges on developer profile..

G.dev/austinedfizer

So, I think I believe to know what is going on. I never brought up the facts your are presenting about what I said. Thanks though!

My sole purpose is let Gemini access your files and devices and or accounts optionally right now.

1 Like

Hi @Austin_Fizer ,

Welcome to the Forum!
Thank you for your feedback. We appreciate you taking the time to share your thoughts with us, and we’ll be filing a feature request.

1 Like

@Austin_Fizer You do not understand the issue as accessing local files isn’t something blocked by AI Studio. It is blocked by the browser you’re using Chrome, Edge, FireFox, Opera, etc. None of these browsers will allow a site to access local files as this would be a huge security problem.

To solve this issue, AI Studio would have to become a desktop application. It could still call Gemini and the other APIs, but it needs to run in a more elevated environment. A browser would not be the proper place for that.

1 Like

@Austin_Fizer I was trying to provide a solution, as this is not the environment for what you are wanting. Yes, it would be totally convenient, I agree, but as @Katje said, it can’t happen.

Google is actively preventing exactly what you are trying to do.

You are using a phone; you are essentially trying to run a heavy-duty industrial lathe inside a phone booth and wondering why there isn’t more room to move.

If Google granted AI Studio the power to reach into your phone’s local folders via a web tab, it would be a massive security regression, not an improvement.

The ‘Fix’ Already Exists: The ‘security’ solution for professional projects is to use a backend (like Firebase) or a repository (like GitHub). These aren’t ‘workarounds’; they are the standardized, secure ways to provide an AI model with access to data without compromising the device’s local integrity.

I was not insulting you; I was showing you how to safely and logically do what you were asking for.

If you need anything, just ask. That’s why we are here; we are learning as we go. Good luck on your project; if you want to talk about what you’re working on, I’m here. Take it easy.

CMD. Proton🖖