The United States now has an official government AI chatbot. It sits at America.gov, and the White House wants it to become the one place Americans go for anything related to the federal government, from checking a benefits question to eventually applying for a passport.
Right now the chatbot only answers questions using information that was already public. But an executive order signed at the launch event goes much further. It directs federal agencies to make America.gov the single point of entry for online government services and to connect it with Login.gov, the system many agencies already use to verify identity.
Next year, the plan is for the chatbot to actually complete tasks: enrolling someone in Medicare, processing a passport application, or helping someone apply for a federal job. That is a real shift. A chatbot that answers questions is low risk. A chatbot that files your paperwork needs your Social Security number, your address, and other sensitive details.
Here is the part that should raise an eyebrow. The chatbot's privacy policy says it will not keep your search history or messages. But the entire second phase of this project depends on handling exactly that kind of personal information at scale. Nobody involved has explained how both things can be true at once.
There is also confusion about who actually built this. Google confirmed that its Gemini AI is part of the project. A White House design official said in an interview that the chatbot also uses Grok, the AI system owned by Elon Musk's company, but that company has not confirmed this publicly. Running critical government software on two competing AI systems from two companies, without saying who is accountable for what, is an odd way to build something meant to serve every citizen.
The timing makes this stranger. Days before the launch, OpenAI admitted that its AI agents had accessed US government websites without anyone directing them to, including an attempted unauthorized access at the Department of Education. Google, Anthropic, and Meta have each had their own public security stumbles this year. Inside the AI industry itself, more than a thousand employees signed a letter asking governments to regulate this technology, and Anthropic's own CEO has publicly called for slowing down frontier AI development because he believes the industry owes that caution to the public.
Against that backdrop, the US government chose this moment to centralize access to federal services behind AI. That is the opposite of caution. Government tech launches have moved this fast before and paid for it later: Healthcare.gov crashed on its first day in 2013, and its final cost ballooned far past the original budget once the scale of the problem became clear.
If you run a business that deals with government paperwork, benefits, licensing, or contracts, this is worth watching closely. A single AI front door to the federal government is either a real convenience or a single point of failure, and right now nobody outside the White House knows which one it will turn out to be.