Democrats demand White House preserve records of AI chatbot use

The demand for transparency

Senior Democrats have formally requested that the White House preserve all records of official exchanges with artificial intelligence chatbots. The move comes as questions arise regarding the extent to which these tools might influence high-level military and foreign policy decisions. Lawmakers Jamie Raskin and Robert Garcia sent a letter stating that any questions submitted by the president or his aides, along with the chatbots' responses, are presidential records.

Under federal law, these records must be kept and protected for future review. The lawmakers are specifically looking for a closed-door briefing scheduled for November 11 to discuss the administration's use of AI. They want to understand the security safeguards currently in place and determine if outside companies have any access to sensitive conversations held within the executive branch.

The request is broad in scope and covers several major commercial AI platforms. The lawmakers are seeking information on the use of OpenAI’s ChatGPT, Anthropic’s Claude, and Google’s Gemini. By demanding these records, the committee aims to ensure that the government is not relying on opaque systems for critical national security matters without proper oversight.

Private AI in the Oval Office

The push for transparency follows reports that President Trump spent significant time using Grok, an AI chatbot owned by Elon Musk. During an Oval Office meeting in December 2025, the president reportedly used the tool to explore potential reactions to the capture of Venezuelan President Nicolas Maduro. The chatbot allegedly suggested that many Venezuelans would welcome Maduro's downfall, shortly before US forces seized him.

While there is currently no evidence that the chatbot actually determined the military operation, the lawmakers are concerned about the reliance on a privately controlled tool. They argue that using a commercial AI for foreign policy creates risks because these systems can be manipulated or provide inaccurate information. The private nature of the software means the government may not have full control over the underlying data or the logic used to generate answers.

The lawmakers are specifically investigating whether the administration has established restrictions on submitting classified information to these chatbots. They want to know how the government handles the logs of these interactions and whether third parties can view the data. The goal is to determine if private corporate interests are being integrated into the decision-making process of the United States government.

Risks of AI intelligence errors

A primary driver for this demand is the potential for AI-generated errors to lead to real-world consequences. Democrats pointed to a reported intelligence error involving a Chinese vessel that was assisted by AI. They are also highlighting allegations that an over-reliance on artificial intelligence contributed to a US strike on an Iranian school. That specific strike resulted in the deaths of more than 100 children.

The lawmakers expressed deep concern that if the military relies on faulty advice from a chatbot, the public has a right to know the details. They are seeking to establish a clear paper trail of how AI is used in tactical and strategic environments. By documenting these interactions, the government can be held accountable for the outcomes of AI-assisted decisions.

The concern is not just about the technology itself but about the lack of a standardized process for its use in sensitive areas. If the White House uses AI to weigh options in a conflict, the logic behind those options must be accessible to oversight bodies. The lawmakers want to ensure that the human element of leadership remains the primary driver of national security policy.

Official response and next steps

In response to the letter, a White House official stated that the administration preserves all required records in accordance with the law. This indicates that the executive branch is aware of its obligations under the Presidential Records Act. However, the specific details regarding the frequency and nature of AI use remain a point of contention for the lawmakers.

The upcoming November 11 briefing will be a critical moment for determining how much the administration is willing to disclose. The committee will likely look for specific protocols regarding the handling of chatbot records and the types of queries allowed. They want to see if there is a clear line between general administrative use and high-stakes military planning.

As the debate over AI in government continues, this situation serves as a test case for how the US handles emerging technology. The focus will remain on whether private AI tools can be used safely in a way that protects national secrets. The outcome of these inquiries could shape the rules for how every federal agency interacts with large language models in the future.