The Trump administration has launched America.gov, an AI chatbot intended as a centralized portal for federal services. Privacy experts warn that caching, log retention (including Cloudflare logs for up to 12 months), and contextual data could expose sensitive information even if message text isn't stored. The White House dismissed critics as "idiots," while researchers recommend not submitting Social Security or Medicare numbers and to verify any chatbot guidance through official agency channels.
White House Dismisses Privacy Fears as It Launches America.gov AI Chatbot, Calls Critics 'Idiots'
Trump launched America.gov to help Americans interact with the federal government, but there could be consequences if you share private details with the chatbot. Kevin Dietsch via Getty Images
The Trump administration on Tuesday launched America.gov, an AI-powered chatbot intended to become a central portal for government information and services. The White House frames the tool as a streamlined way for Americans to access federal resources — including passport renewal now and planned Medicare enrollment later this year — but privacy and accuracy concerns have drawn sharp criticism from security researchers and policy analysts.
What America.gov Promises
The administration describes America.gov as a "single point of entry for Americans to access covered services online." The site warns users not to share Social Security or Medicare numbers and states that "messages aren't saved by America.gov, and your chat disappears when you leave. Some answers may be cached for up to two hours." Officials say the chatbot is powered by the Gemini and Grok models.
Privacy And Data-Retention Concerns
Experts caution that seemingly innocuous context — such as age, location, medical conditions, or employment status — can combine to identify individuals and expose them to discrimination, fraud, or targeting. America.gov's privacy policy also notes that Cloudflare security logs, which may include IP addresses and timestamps, could be retained for up to 12 months. Researchers say that while query text may not be stored, visit metadata is still data and can reveal user behavior.
Experts' Warnings
"I would absolutely advise against submitting sensitive information or using it to apply for any benefits when those features are added," said Philip Fields, a cybersecurity researcher and former FBI counterintelligence analyst. "The risk of wrong answers is too high and the government has a mixed record on protecting Americans' privacy."
Nicole Alvarez, senior technology policy analyst at the Center for American Progress, highlighted how descriptive personal narratives can effectively identify people even when specific identifiers like Social Security numbers are omitted.
White House Response
When asked to respond to privacy critiques, a White House official told HuffPost: "These researchers are idiots. No data or messages will be stored by America.gov, nor will they be shared with Gemini or Grok for LLM training." The administration maintains chats will not be used to train the underlying models.
Past Incidents And Trust
Researchers pointed to earlier reporting that the National Design Studio — the contractor behind America.gov — had previously installed visitor-tracking software on federal websites without clear public disclosure; that code was removed after press reports. Advocacy groups also cite prior cases in which government-held personal data was repurposed across agencies without individuals' consent, underscoring concerns about "secondary data abuse."
Accuracy Risks: AI "Hallucinations"
Independent experts warn that large language models can produce incorrect or fabricated information — often called "hallucinations" — whether the backend is OpenAI's ChatGPT, xAI's Grok, or Google's Gemini. For critical processes like benefits applications, inaccurate guidance could lead to denied claims or other consequences for applicants.
Practical Advice
Security and policy experts recommend: do not submit Social Security, Medicare, or other sensitive identifiers via the chatbot; verify any guidance with agency hotlines or official websites; and avoid relying solely on AI-generated answers for applications or decisions that affect benefits or legal status.
Bottom line: America.gov aims to simplify access to federal services, but caching, log retention policies, past contractor practices, and the risk of AI errors mean users should proceed with caution and protect sensitive personal information.
Help us improve.
