Gemini API Logging Is Now On by Default: How to Turn It Off

Updated November 1, 2026. On October 26, 2026, Google changed a default in the Gemini API: new paid-tier projects now store Generate Content API requests by default (store=true), matching the Interactions API (Gemini API changelog). That covers generateContent, streamGenerateContent and batchGenerateContent. The stored logs contain your full prompts and responses and are visible in Google AI Studio. If your team sends customer or internal data to the API, the first thing to check is when each project was created.

Short answer: projects created on or after Oct 26 log by default

Here is how the defaults now break down, based on Google's Logs and datasets documentation.

You can opt out in two ways: turn off Generate Content API logging for the whole project in the Settings panel of the AI Studio Logs page, or pass store=false on individual requests. The request-level setting takes precedence over the project setting.

Are stored logs used to train Google's models?

Not by default. Google's logs policy says that because logging is only available for billing-enabled projects, prompts and responses in logs are not used for product improvement or development by default.

Your data can be used for training only if you opt in: by saving logs into a dataset and sharing it with Google, or by submitting feedback (thumbs up/down and comments) on logs. Shared datasets are processed under the data-use terms for Unpaid Services, which allow model training. Google explicitly warns not to include personal, sensitive or confidential information, and notes that sharing options depend on your location.

So "stored" does not mean "trained on." But it does mean the contents of your prompts now sit in your project's log viewer, readable by anyone with access to that project. In practice, that access list is the part most teams have never reviewed.

How long logs are kept: 55 days by default

Logs expire and are marked for deletion after a default window of 55 days. You can shorten it per project to 7, 14, 28 or 55 days. Logs saved into a dataset do not expire, so a debugging dataset can quietly become the longest-lived copy of your data.

These project logs are separate from logs Google keeps solely for abuse monitoring. The settings described here control the project logs only.

How to turn logging off

Project level (AI Studio)

1. Open the Logs and Datasets page in Google AI Studio.
2. Select the project.
3. In the Settings panel, switch off Generate Content API logging. Generate Content API and Interactions API logging are toggled independently.

Warning: turning off Interactions API logging stops the API from automatically storing and retrieving conversation history unless you override it per request. Check this before switching it off if your app relies on server-side state.

Request level (store=false)

If only some calls carry personal data, set storage per request: config={'store': False} in Python or config: { store: false } in JavaScript. Setting store on Generate Content API requests requires these SDK versions or later:

Upgrade the SDK at the same time you add the flag. Calls made through the OpenAI-compatible endpoints are also covered by logging, according to the documentation.

What is never logged

• Imagen and Veo models
• Gemini embedding models
• Gemini Robotics models
• Inputs containing videos, GIFs or PDFs
• Public Preview Agents in the Gemini API

Also note: if you delete every API key in a project (for example after a key leak), you lose visibility of that project's logs until a new key exists. The logs are hidden, not deleted.

A 5-point checklist for small teams

A common pattern we see is a mix: last year's production project does not log, while a test project created last week logs everything. Go through these in order.

1. Creation date: list every project created on or after Oct 26, including ones set up by contractors.
2. Data in prompts: do prompts include customer names, contact details, contracts or confidential figures? If so, turn logging off or use store=false for those calls.
3. Retention: if you keep logging, is 55 days necessary? 7 or 14 days is usually enough for debugging.
4. Who can read logs: anyone with project access can. Remove former staff and old contractor accounts.
5. Dataset sharing: shared datasets can be used for training. Write "we do not share datasets" into your internal AI policy.

Teams that build their own AI tools are the ones most likely to miss default changes like this. Mihata helps companies design custom AI systems that handle internal data, including logging and access controls. If you would like a second pair of eyes on your setup, feel free to take a look.

Summary

Since October 26, 2026, new paid-tier Gemini API projects store Generate Content API requests by default. Stored logs are not used for training by default, but full prompts stay in your project's log viewer for up to 55 days. If you send personal or confidential data, switch logging off in AI Studio or use store=false with an updated SDK, and set a retention window and access list.