President Donald Trump unveiled America.gov on Sept. 29, calling it a transformative tool that would finally put government services within reach of ordinary Americans. “The federal government no longer stands in your way,” Trump said at the launch event, “and it stands only at your service.” The AI-powered site aims to make federal benefits and services easier to access. People can use it to change their names or addresses, check eligibility for student loans or Medicare, and navigate bureaucratic systems that have long frustrated citizens seeking help.
But as the government puts sensitive personal data in the hands of AI systems, security researchers are raising urgent questions about whether the technology can handle that responsibility. The gap between the administration’s confidence and expert caution follows a familiar pattern: bold promises about innovation, paired with insufficient safeguards for the people affected.
What America.gov Is Supposed to Do
The site’s training data included 29,000 U.S. government websites and 800 million pages. Joe Gebbia, the chief design officer of the United States and architect of the portal, described it in an interview with Fox News as a “concierge to navigate the government.” “Don’t think of it as a website; it’s more of a conversation that you start with the government to try to get access to the answers you need,” he said. Trump said the site would handle applications and renewals on users’ behalf by early 2027, and even help people apply for passports.
The General Services Administration, the Office of Management and Budget, and the National Design Studio—a White House initiative established by Trump in 2025—co-created and run the system. An executive order directs federal agencies to integrate their services with the portal, extending its reach across government.
On the surface, the promise is compelling. Each year, millions of Americans struggle to access benefits they qualify for, often because applications make little sense or the right agency is hard to find. A tool that genuinely simplifies the process could make a real difference in people’s lives—helping seniors enroll in Medicare, students understand loan options, or families access support programs they’ve earned.
The Security Problem No One’s Solved
The same design that makes the tool useful also creates what security experts call an “attack surface”—a set of vulnerabilities bad actors could exploit. Jessica Ji, a senior research analyst at Georgetown’s Center for Security and Emerging Technology (CSET), said AI chatbots have inherent weaknesses that attackers may find before anyone else can detect them.
“When an AI system is empowered to handle private data and act for users, I think you are increasing the attack surface quite a bit,” Ji told USA TODAY. At the launch, Trump claimed the site was “built from the ground up with your privacy and security totally in mind,” and said it wouldn’t collect or store personal information. Ji was direct about the limits of that assurance: “It’s basically impossible to tell what those weaknesses are before they are discovered.”
Early tests have already found problems. Reddit users posted examples of the chatbot giving nonsensical answers to simple requests like “Play Minecraft.” USA TODAY tried the same request and received a similar random response. The site also contains spelling errors, unproven claims, and confusing answers to user questions—red flags suggesting officials didn’t adequately test the technology before trusting it with sensitive government data.
Ji acknowledged that the portal could serve a useful purpose as a landing page to help people navigate government systems. That use, though, doesn’t erase the underlying risk. Once the system begins acting on users’ behalf—filling out applications, processing renewals, applying for documents—it will handle the very personal information that makes an attractive target.
Why This Matters
The moment captures a broader tension in the way government adopts new technology. The push to simplify bureaucracy is genuine and necessary: too many eligible Americans miss out on benefits because the system is deliberately or accidentally incomprehensible. But rushing to deploy AI without adequate security checks puts ordinary people at risk. Those most likely to benefit from a streamlined government portal—seniors, low-income families, people with limited digital literacy—are also least able to recover if someone compromises their personal information. The gap between officials’ promises and independent experts’ assessments of what’s actually secure isn’t a minor technical detail; it’s a question of whether the people who most need government help are being protected or exposed.