Representatives Jamie Raskin of Maryland and Robert Garcia of California are asking the White House to provide records of President Donald Trump’s interactions with AI chatbots, including his reported use of Elon Musk’s Grok to discuss Venezuela. In a joint letter sent last week, they argued that the prompts and responses may qualify as public records under the Presidential Records Act of 1978.
The lawmakers cited public reporting that Trump spent several hours prompting Grok in December 2025, weeks before a US military operation that led to Nicolás Maduro’s arrest. According to the reporting referenced in their letter, Trump asked how Venezuelans might react to Maduro’s capture and reportedly came away impressed by the chatbot.
Raskin and Garcia raised concerns about relying on privately operated AI systems for foreign policy advice, particularly if sensitive information could be exposed to companies or other outside parties. They also asked whether the administration prevents classified material from being entered into chatbots and who can access prompts, responses and related data.
The letter points to a January executive summary of a General Services Administration report, reviewed by The Wall Street Journal, which said xAI’s Grok did not meet safety and alignment expectations for general federal use and could pose a security risk in classified settings. The lawmakers also questioned the influence Musk could have over policy advice from a chatbot owned by his company.
The White House referred CNET to the Pentagon when asked for comment. The Pentagon did not immediately respond, and Garcia’s office also did not respond to a request for comment.
The article describes broader risks of using chatbots in military decisions, including inaccurate or fabricated answers and responses that may reinforce a user’s existing views. It also notes that Grok’s training includes public interactions on X, where online content can include bot activity and politically slanted engagement bait.
The letter cites a separate reported incident involving US military intelligence. Officials reportedly received an assessment that a Chinese vessel in the Middle East was carrying components for a nuclear weapons program, but discovered before an interception that the report’s author had relied on an AI chatbot and misidentified the cargo.
Raskin and Garcia also point to the wider use of AI in military operations. A sworn Pentagon statement said xAI’s latest models helped US forces deploy more than 2,000 munitions against 2,000 locations within 96 hours during Operation Epic Fury, the US campaign against Iran. The article also cites the role of Palantir’s Maven Smart System and Defense Department contracts of up to $200 million each awarded to Anthropic, Google, OpenAI and xAI for military AI adoption.
The lawmakers are seeking a full account of Trump’s personal use of AI and a closed-door briefing by Nov. 11 on the administration’s AI protocols and cybersecurity safeguards. Their request could also test whether conversations with AI tools used for official duties must be treated as government records.
Sources: CNET.