Google’s new Prompt API, designed to enable web pages to interact directly with local AI models in Chromium-based browsers, faces criticism from Mozilla. The organization warns that this move could fragment web standards and impose restrictive usage policies.

  • Prompt API enables direct AI inference in browsers locally
  • Mozilla warns of risks to web platform neutrality and interoperability
  • Google’s model-specific policies could limit developer freedom

Product angle

The Prompt API introduced by Google is an emerging browser feature that standardizes communication between web pages and a local AI model embedded within the Chromium environment. It allows for natural language instructions to be sent directly to a relatively lightweight AI, the Gemini Nano model, providing benefits such as faster response times and offline usage. This approach contrasts with traditional methods where AI interactions occur through remote cloud services requiring network access and API keys.

While the technology showcases potential improvements in cost-effectiveness and user privacy through local AI inference, it also binds the development ecosystem closely to Google's AI model and its unique usage policies. This raises concerns that web developers may find themselves dependent on a proprietary AI infrastructure and compliance requirements. These constraints challenge the typical openness and neutrality expected of web platform APIs.

Advertising
Reserved for inline-leaderboard

Best for / avoid if

This API may appeal to developers seeking a fast, integrated AI inference experience without relying on cloud services, particularly in environments where offline or low-latency operation is critical. Those looking to experiment with embedded language models within browsers may find the Prompt API intriguing for initial exploration or prototyping.

However, buyers and developers aiming to build truly open, interoperable web applications without vendor lock-in should approach with caution. Mozilla’s critique highlights that reliance on a single vendor’s AI model, together with policies governing how the API can be used, could limit flexibility and cross-browser compatibility. Organizations valuing adherence to open standards and multiple AI backend options may prefer to avoid adopting solutions dependent on this API.

Pricing and alternatives to check

Specific pricing details for the Prompt API itself are not publicly disclosed, as this is a browser-integrated capability rather than a separately monetized cloud service. However, usage requires significant local storage capacity—several gigabytes for the AI model download—which might factor into operational costs indirectly, such as device requirements and maintenance.

Alternatives to the Prompt API include leveraging established cloud-hosted AI services accessible through traditional REST APIs, which provide broader model choices and vendor neutrality, albeit often with latency and usage fees. Other browser-based options include running AI models via JavaScript runtimes, WebAssembly, or WebGPU, which developers can tailor more freely without depending on a single vendor’s policies.

Source assisted: This briefing began from a discovered source item from The Register Headlines. Open the original source.
Review disclosure: Review-watch pages are buyer briefings unless clearly labelled as hands-on SignalDesk reviews. Affiliate, sponsor or free-access relationships should be disclosed on the page. Read the review methodology.
How SignalDesk reports: feeds and outside sources are used for discovery. Public briefings are edited to add context, buyer relevance and attribution before they are published. Read the standards

Related briefings