Back to Home
AI Platforms4 min readSource: Ars Technica

Google Gives Schools a Choice on Gemini, but Not on AI Search

Google reportedly lets schools opt out of Gemini while keeping Search, yet AI-powered search results stay mandatory. What this means for API builders.

What Google is actually offering

According to Ars Technica, Google is telling schools that they can continue using Search while declining to adopt Gemini as a separate product. That sounds like flexibility, and in a narrow contractual sense it is: an institution can say no to one branded AI tool without losing the search engine its students and staff rely on.

The catch, as the reporting frames it, is that AI-generated elements inside Search itself are described as non-negotiable. In other words, the opt-out covers the assistant-style product, not the model-driven layer that increasingly determines what a search results page looks like.

Why the distinction is more than semantics

For a school administrator, 'we turned off Gemini' may sound like a clean boundary. But if AI summaries, generated overviews, or ranked AI answers appear inside the default search experience, students are still interacting with model output regardless of which toggle was flipped.

This is the recurring pattern in platform AI rollouts: the visible product can be optional while the underlying behavior becomes the default. Buyers evaluating any AI-heavy service should ask which layer the opt-out actually governs, not just which brand name disappears from the menu.

The vendor-neutral lesson for developers

If you build on top of a search or assistant API, treat product-level controls and model-level behavior as separate questions. A provider may let you disable a named feature while still routing your requests through the same generative pipeline. Read the documentation for what changes in the response payload, not only for what changes in the billing line item.

It also reinforces a design habit worth keeping: keep your own abstraction layer between your application and any single provider. If the AI layer inside a core service shifts, an internal interface lets you swap models, adjust prompts, or add filtering without rewriting the whole product.

Finally, institutions and enterprises with compliance obligations should document which AI features are active by default. 'We opted out' is only defensible if you can show precisely what you opted out of.

What is still unclear

The reporting does not spell out how granular the controls are, whether AI overviews can be disabled at the domain level, or how Google verifies institutional settings. Those details would determine whether the offered choice has real operational weight.

Until then, the safe reading is that Google is preserving access to Search for education customers while declining to make its AI search layer optional. That is a meaningful concession on one axis and a firm line on another.

FAQ

Can schools keep Google Search without Gemini?
Ars Technica reports that schools can retain Search while skipping Gemini as a product. However, AI-generated elements within Search are described as non-negotiable, so the opt-out does not appear to cover the AI layer itself.
Does this affect developers using Google AI APIs?
Not directly, since the reporting concerns education customers and the Search product. The broader lesson is to distinguish product-level opt-outs from model-level defaults when evaluating any AI platform.
How should teams handle AI features they cannot disable?
Build an abstraction layer, log which AI features are active, and document defaults for compliance reviews. This keeps you able to adapt if a provider changes what is on by default.

References

Call these models through one gateway

Admesh provides a unified OpenAI-compatible API: one Base URL and one key for Claude, GPT, DeepSeek and more. New accounts get trial credit that never expires.