Google shipped its Prompt API in Chrome 148 on May 5, embedding a JavaScript interface to its Gemini Nano AI model directly into the browser despite formal opposition from Mozilla, Apple's WebKit team, the W3C Technical Architecture Group, Microsoft, and a Chrome engineer who helped approve the feature. The API lets any webpage send natural language instructions to Gemini Nano for text generation, summarization, classification, image captioning, and audio transcription using a 4.27 GB model Chrome downloaded to eligible users' machines without prompting, with the model re-downloading itself even after manual deletion unless users toggle off the local AI in System settings. Mozilla called the API a feature with “severe negative consequences to the interoperability, updatability, and neutrality of the web platform,” and critics also flagged that developers must accept Google's Generative AI Prohibited Uses Policy (which prohibits content Google deems “disturbing” or classifies as misinformation) as a precondition to using a browser API, a first for the web platform that historically has been governed by open specifications rather than a single vendor's usage policy.
