Trump launches America.gov AI chatbot for federal services

By Splitfeed EditorialLast updated Sep 30, 2026, 23:30 UTC

On September 29, 2026, the Trump administration launched America.gov, an AI-powered portal designed to serve as a single entry point for federal government services through a chatbot interface. Trump signed an executive order requiring federal agencies to integrate with the site. White House official Gebbia confirmed the chatbot runs on multiple AI models, including Google Gemini and Grok, the latter built by Elon Musk's xAI company. The site was built by alumni of the Department of Government Efficiency, known as DOGE.

Within hours of going live, the chatbot drew attention for contradicting claims Trump made during his own unveiling remarks, including on the 2020 election. Observers noted that the chatbot's answers on politically sensitive topics shifted during and after the event. Beyond the immediate controversy over those answers, the site faces a more structural credibility question: services such as passport applications, Medicare enrollment, and federal job listings are not expected to be functional until 2027, meaning the launch was largely a demonstration rather than a working product. Experts have flagged that the large language models powering the chatbot remain prone to hallucinations, raising concerns about accuracy for users seeking official government information.

The pro-AI camp, as reflected by accounts like tominco.bsky.social, frames the criticism as reflexive AI skepticism and treats the launch as straightforward modernization progress. Critics in the anti-AI camp are more pointed. Accounts including Angry Donkey News raised alarms about the White House app covertly collecting excessive user data and sharing it with third-party trackers while bypassing federal security rules. Mark Chadbourn described the accompanying regulatory framing as a non-agreement designed to create the appearance of oversight while leaving commercial AI development unchecked, and Olga Nesterova questioned the logic of a White House accord on self-regulating superintelligence. The middle ground camp found a specific irony worth noting: Santiago Mayer and Quinn Yeargain both pointed out that the chatbot actually answered accurately on questions that Trump administration officials have repeatedly refused to address clearly in other contexts. The Washington Sun noted the gap between the promotional fanfare and the 2027 timeline for actual functionality, and Tuffy offered the view that a government AI portal of this kind would likely have arrived regardless of who was in office.

The arguments most likely to be tested in the near term concern data privacy practices, whether the chatbot's outputs on sensitive topics stabilize or continue to shift, and whether federal agencies meet the integration requirements set out in the executive order before the promised 2027 service launch date.