A cloud setting can change for your next project without changing the one you already run. Google’s October 26, 2026 Gemini release notes give a concrete example: new paid-tier projects using the Generate Content application programming interface (API), the connection software uses to send requests to Gemini, now store requests by default. Existing projects retain their current setting. For learners, the important distinction is between a default you inherit and a requirement you cannot change.
Which projects changed?
The announcement covers generateContent, streamGenerateContent and batchGenerateContent. Google’s logging documentation says projects created on or after October 26 default to store=true. Older projects remain at their existing setting, normally store=false unless logging was previously enabled. Storage for these developer logs is available on the Gemini API paid tier. Do not extend this announcement to every Google AI product or assume it describes another platform’s settings.
Google says a log entry can show the full prompt and response, along with context from previous turns. That makes the setting useful to inspect before putting sensitive material into a learning application. A request that successfully receives an answer does not, by itself, tell you which logging choice applied.
Two controls, with a clear order
The same documentation describes a project-level control in the Settings panel of the AI Studio Logs dashboard. Generate Content and Interactions logging can be toggled independently. An individual request’s store property takes precedence over that project-level setting. Setting store=false is therefore a request-specific opt-out, not a change to every later request from the project.
There is also a client-library detail. Google documents minimum library versions for setting this property on Generate Content requests: Python’s google-genai needs version 2.33.0 or later, for example. If you later build a lab, check the documented version for your chosen language rather than assuming an old example supports the new setting.
Work through a paper example
Imagine two paid projects: one created before October 26 with logging never enabled, and one created after the change. This is a learning scenario, not a test performed against a live account. Using Google’s documented precedence rules, explain these outcomes:
- The older project keeps its existing default; its creation date does not force the new setting onto it.
- A request from the newer project inherits storage enabled if neither the project preference nor the request overrides it.
- That newer project can send a particular request with
store=false, overriding the project default for that request.
For each case, write down the project setting, the request setting and the expected logging behavior separately. If you later run the exercise in an authorized lab, use harmless sample text and record what you actually observe. Do not substitute a prediction for a test result.
Storage is not the whole data-policy decision
The logging guide says logs expire and are marked for deletion after a default 55-day retention window; the available project retention choices are 7, 14, 28 or 55 days. Logs saved into datasets do not expire under that rule. The guide also describes choosing to share datasets with Google. Keep logging, retention, dataset creation and sharing as separate questions rather than treating one switch as the answer to all of them.
For Generative AI Leader study, use this case alongside the Nex Arc Generative AI Leader study guide as practice in asking precise data-governance questions. This is a current product example, not a claim that a particular dashboard control or library version will appear on the exam. The transferable habit is to identify the product, project and override before deciding what a default means.