The White House Launched a Gemini and Grok Powered Government Chatbot

The Trump administration unveiled America.gov, an AI chatbot built on Google's Gemini and xAI's Grok meant to serve as a single entry point across roughly 29,000 federal websites, with a second phase planned for 2027 letting users apply for benefits like Social Security directly through the chat interface.

Sep 30, 2026
The White House Launched a Gemini and Grok Powered Government Chatbot
White House Just Launched a Gemini and Grok Powered Government Chatbot

Navigating federal websites just got a conversational front door, and two rival AI labs are the ones answering the questions behind it. The Trump administration launched America.gov on Tuesday at a day-long event in Washington, with Chief Design Officer Joe Gebbia confirming to CNBC that the chatbot runs on Google's Gemini and xAI's Grok, drawing answers from across roughly 29,000 federal websites.

Built by the White House's National Design Studio and operated through the General Services Administration, the site is designed to replace manual searches across a famously fragmented federal web presence with a single conversational interface. A few details define how it's meant to work and where it's headed:

  • Trump signed an executive order the same day mandating that public-facing federal services serving more than 100,000 users annually integrate directly with America.gov
  • A planned "phase two" in early 2027 would let Americans apply for and track benefits like Social Security directly inside the chatbot, including uploading documents such as marriage certificates
  • Per its privacy policy, the chatbot won't save search history or messages, and doesn't require an account to use

The pairing of Gemini and Grok is notable on its own — two models from companies with very different relationships to the administration, and notably not the single in-house model many expected.

As TechCrunch's analysis points out, the stakes here aren't abstract: people may rely on America.gov for food stamp applications, visa renewals, or tax filings, where an AI error could mean "missed deadlines, denied benefits, or penalties" rather than just an unhelpful answer.

The launch puts two commercial AI models in a direct, high-consequence role serving the public — and unlike a company chatbot, there's no simple option for a frustrated user to just call a different provider if the answer turns out to be wrong.