The Prompt API's standards problem

Google has been quietly shipping Gemini Nano to Chrome users as a roughly 4GB download — no consent prompt, no opt-in explanation. Remove it and Chrome fetches it again. The model is bundled with the browser but not integrated into it, which raises a question: why is a standalone product being treated as part of Chrome itself?

The Prompt API has been a known quantity for over a year, with published explainers and an intent-to-prototype filed with the Blink team. What's newly contentious is Mozilla's formal opposition, which centers on the terms Chrome attaches to API usage. Developers must acknowledge Google's Generative AI Prohibited Uses Policy before using the Prompt API, and portions of that policy reach beyond what the law requires. Mozilla flags restrictions on sexually explicit content and on misinformation related to governmental or democratic processes as examples of conditions that go further than legal mandates.

"This seems like a bad direction for an API on the web platform, and sets a worrying precedent for more APIs that have UA-specific rules around usage."

That precedent is not hypothetical — the API has already shipped. Google's standards participation, in Mat Marquis's framing, resembles less a collaborative process than a bear's approach to camping. The company has cited developer sentiment in support of the API while also citing venues where that sentiment is skeptical or absent.

Browser APIs versus Web APIs

The distinction matters beyond this single API. Browser choice has long been constrained in ways Alex Russell and others have documented, and the Prompt API sharpens an older problem: not every API exposed by a browser is a Web API. When a browser vendor ships an API with usage rules tied to its own generative AI policy, the line between platform feature and vendor product blurs.

Chrome browser settings with system tab open showing disabled on-device AI option.
## Broader coverage

For those following the story from other angles: