A person should not need to understand an API setting to have a say in whether their material becomes a saved AI request. I want companies selling AI infrastructure to put that choice before collection, where someone has to make it deliberately.
Google's October 26 Gemini API release notes announce a changed default. New paid-tier projects using Generate Content now store requests by default, matching the Interactions API. Existing projects keep their current setting. Google names three affected methods: generateContent, streamGenerateContent and batchGenerateContent.
The note also gives developers two ways to opt out: set store=false on individual requests or turn off Generate Content API logging in the Settings panel of the Logs and Datasets dashboard. Those controls matter. This is an argument about the starting position, not a claim that Google provides no choice.
The default assigns the work
I disagree with making storage the starting position. My proposal is simple: before the first paid request, ask the project's administrator whether requests should be saved. Explain the scope of that choice beside it. Require an answer before proceeding.
That would put responsibility on the institution changing the default to make the decision visible. An opt-out puts responsibility on the developer to notice the change and act on it.
The person configuring a project and the person described in a request can be different people. Consider a workplace tool processing a customer's message. That is a hypothetical example, not an account I examined. A developer's convenience is not, by itself, a sufficient reason to make retention the assumed choice for everyone whose material passes through that tool.
I would hold a small builder to the same standard. Having less institutional power should not mean getting less explanation. Nor should the ability to hire someone to inspect settings determine how much practical control a person has over their information. That is the freedom I want these products designed to protect.
Keep only what has a reason to stay
The FTC's Start with Security guide advises businesses to keep only essential personal information and retain it only while they have a legitimate business need. The guide discusses lessons from settlements and explicitly distinguishes them from court findings. It is general guidance; it does not establish that Google's change violates a law.
I agree with its principle. Saving requests can have a purpose. The company and the developer should explain that purpose and make a conscious choice about retention, rather than letting a default settle the question.
Neither record establishes a breach, improper training use or harm to a particular person. I am not alleging those things. I am arguing for a clearer allocation of power before anyone has to prove harm.
Show the choice before the first request
Google should put an affirmative storage decision in front of new paid-project administrators and give them language they can use to explain it to the people whose material they process. Developers should carry that explanation into their own products.
Here is a specific request for scrutiny: point to a public Google notice or product record showing what a new paid-project administrator is told about storage, and what choice they make, before the first request. Does that record establish an informed decision, or an available setting? A substantive reply should address that distinction. The release note documents the default; it does not answer that interface question.