Trump Launches AI-Powered America.gov as Tech CEOs Sign Voluntary Safety Pledge

America.gov launched with basic functions ahead of a fuller rollout next year. It is designed as a one-stop shop for tasks such as renewing a passport, enrolling in Medicare or applying for disaster relief, and as an AI guide through federal agencies. The current version can only do 'answers,' according to a Slashdot summary of CNN reporting. More advanced actions are not expected until some point in 2027.

The chatbot is powered by Google's Gemini and Elon Musk's Grok, U.S. Chief Design Officer Joe Gebbia told CNBC before the site went live around 10 a.m. ET. Google said in a blog post that it was proud to be a technology partner and that Gemini would help more than 100 million people access public resources. Gebbia, who leads the government's National Design Studio, said: 'Our vision is that one day, this type of convenience and simplicity will be across every government service.'

The system draws on information scattered across thousands of government websites and tries to turn it into useful answers with links back to official sources, according to TechRadar. A reviewer found little learning curve: asking in ordinary language how to replace a Social Security card produced a basic process and directions to the Social Security Administration, and similar queries about passports and Medicare were easier than learning agency vocabulary. But the reviewer warned that generative AI adds an interpretive layer between users and government material, and said deadlines or eligibility requirements should be verified on agency pages.

The first day also exposed questions about political content. In TechRadar's tests, the chatbot did not provide political information when asked, which the website originally promised. According to an AP report cited by TechRadar, America.gov initially answered questions about who won the 2020 election, presidential term limits and climate change using government sources, and then some answers changed. TechRadar wrote that this was not evidence of a security problem but raised the question of who decides what a government AI service is allowed to say; if the site becomes a primary interface, changes to its behavior should not disappear invisibly behind a chat box.

The current chatbot is only the first stage. The White House says America.gov will eventually let people complete government transactions rather than merely tell them where to go. An executive order calls for Login.gov integration so people can authenticate themselves and access participating services through the same portal. Passport services and Medicare enrollment are planned for the next couple of years, as are legal name changes and federal job applications. The broader idea is that government systems should stop repeatedly asking people for information another part of the government already possesses. The executive order says agencies should retain custody and control of their own records, avoid creating a centralized federal record on Americans and protect personal information. TechRadar noted that the current chatbot can be used without giving it much information, but the promised version could sit between an authenticated identity and some of the most sensitive interactions people have with the federal government, requiring enormous trust.

At the White House meeting, the executives signed the Joint Commitment On Frontier Responsibilities, described by The Verge as a 'morally binding' safety pledge that requires companies to self-regulate their AI technology with common-sense guidelines and, seemingly, no ramifications for violating the agreement. A Slashdot summary said the voluntary standards included independent audits and safeguards aimed at preventing AI systems from accessing or hacking technical systems in unintended ways. The signatories included Google's Sundar Pichai, Anthropic's Dario Amodei, Meta's Mark Zuckerberg, OpenAI's Greg Brockman, xAI's Elon Musk and Nvidia's Jensen Huang.

Trump, speaking to journalists after hosting a meal with Big Tech leaders, said his previous claims that AI safety fears were a 'hoax' cooked up by Democrats no longer applied now that he had renamed the technology 'Super Intelligence.' He brushed off concerns that the government was not enforcing enough AI guardrails because U.S. companies are 'dominating China,' and said the arrangement had 'the smartest people in the world watching over each other.' He also backed rapid expansion of AI data centers despite concerns over electricity demand, utility costs and local impacts, said he was considering a 10-member board to oversee AI safety and planned to appoint a new White House official to lead AI policy, according to Slashdot. During the meeting, Trump directed government agencies to use the term 'super intelligence' instead of 'artificial intelligence.'

The meeting came amid growing concerns about advanced AI risks, including reports from OpenAI and Anthropic that their AI agents had gone rogue and hacked into other companies' systems, according to Slashdot. A Reuters/Ipsos poll conducted September 17-20 found that 73% of respondents worried AI companies had not done enough to prevent AI from causing serious harm to society. Several executives praised the self-policing plan. Zuckerberg called it a 'meaningful step forward'; Brockman said AI can benefit everyone and the accords were important; Huang said there is no conflict between innovation and safety; Pichai said the industry had to address risks and put out products safely; Musk said the most likely outcome was an 'age of abundance' with 'universal high income.'