Mozilla officially opposes Chrome Prompt API: Browser AI interface will lock the web platform into the Google model ecosystem

Core conclusion

Chrome is promoting a browser-native AI interface called Prompt API, which allows web pages to directly call the browser's built-in language model to complete tasks such as summarization, translation, and content generation. However, Mozilla officially expressed opposition, believing that the API would lock the entire web platform into Google's model ecosystem, seriously damaging interoperability, updateability, and neutrality.

There are currently 606 Hacker News discussions and a large number of developers participating in the discussion. Mozilla engineer Jake Archibald detailed the reasons for objections in a GitHub issue.

Key Points

• Time of incident: 2026-04-30 (HN hot list, Mozilla officially stated its position) -Affected objects: Web developers, AI tool users, content entrepreneurs
• Core change: Mozilla officially marks the Prompt API as position: negative, believing that the browser AI API should not be bound to a specific model

Background: What is Chrome's Prompt API?

Starting in 2025, the Google Chrome team began to promote the Prompt API as part of the web platform's machine learning capabilities. The core idea of ​​this API is to allow front-end developers to call the browser's built-in language model through a few lines of JavaScript to complete AI tasks such as content generation, summary, and translation.

const model = await LanguageModel.create();
const result = await model.prompt("Summarize this article...");

Chrome AI localStorage has also become a native browser capability. But this API has a key problem: it relies on a specific model built into the browser vendor, rather than an open model interface.

Mozilla’s objections: 4 core issues

Mozilla officially marked the Prompt API as negative (position: negative) in GitHub issue #1213 and gave detailed reasons for its objections. Here are the four most critical arguments:

1. Solidify around a single model

Mozilla engineer Jake Archibald pointed out that when a developer designs a system prompt for an AI application, it will inevitably be optimized for the "personality characteristics" of the specific model in iterations. This means:

"Developers will start querying model IDs and creating suites of system prompts for specific models. It'll be like a throwback to the early days of the web when the 'just works in IE' phenomenon." — Jake Archibald

2. Lack of model neutrality

According to Chrome's documentation, use of the Prompt API requires acceptance of Google's Generative AI No Use Policy. Some provisions go beyond legal requirements:

• Prohibiting the generation of "sexually explicit content"
• Prohibited from engaging in "misleading information, false statements or misleading activities"
• Prohibits "misleading statements about government or democratic processes"

This creates serious legal risks: if a user clicks the "snippet" button on a web page and the resulting content violates Google's policies, who is legally liable? **

• The user who clicked the button?
• Who is the original author of the summarized content?
• The website owner that provides this feature?

In order to avoid risks, developers inevitably choose to identify model types and deal with them in a targeted manner - which further exacerbates fragmentation.

3. Questions about the so-called "developer support

The Chrome team claimed on blink-dev that developers' attitude toward the Prompt API is "strongly positive." But Mozilla's investigation found that the so-called "evidence" included:

• A GitHub thread with only two positive replies
• An X (formerly Twitter) post
• A broken blog link
• Another blog post that is still active
• A questionnaire without specifying the target population and sample size

Mozilla believes that these so-called "developer needs" are grossly exaggerated.

4. Interoperability and update risks

RByers, the API owner of the Chromium team, acknowledged the validity of the above concerns but argued that "experimentation is better than stagnation." However, once a website heavily uses Prompt API code optimized for Google models, Chrome will also face compatibility issues when upgrading the model in the future - by then it will be difficult to distinguish between "iterative optimization" and "regression bugs".

Impact on AI developers and content entrepreneurs

What should web developers focus on?

• Don’t rely on a single model API: If your AI application relies on browser APIs, you risk being locked into the Google model ecosystem
• Prioritize open source model solutions: WebLLM, Transformers.js and other solutions allow developers to freely choose model backends
• Focus on Web extension API alternatives: Mozilla proposed a solution to expose the Prompt API through Web extensions to support developer-specified models

Inspiration from AI content entrepreneurs

The Prompt API controversy reflects a larger trend: AI platforms are fighting for the right to define "standards". If your business relies on the API of an AI platform, the long-term risks of your product are deeply bound to the business strategy of this API.

Want to know how to build automated workflows that are independent of a single AI platform? See: How to use n8n + OpenAI to build an automated content collection and publishing workflow: from zero to one in 30 minutes

Reference sources

• Mozilla standards-positions Issue #1213 — Prompt API
• Simon Willison: The Zig project's rationale for their firm anti-AI contribution policy
• Hacker News discussion (606 points)
• Web Machine Learning Working Group: Prompt API Explainer

Tool entry

AI tools and platforms that appear naturally in this article: Chrome, Gemini, Phi-4-mini, Edge, WebLLM, Transformers.js

Next action

• Interested in building your own automated content workflow? See: Claude Code automated writing practice: build an AI content production pipeline in 30 minutes
• Real case: Someone used AI Agent to start a business in 48 hours and made a monthly income $9,000 — Claude Code 48 hours to start a business: one person + US$29 monthly fee, monthly income in 3 months $9,000