Congressional Democrats are pushing to access records related to President Donald Trump and other White House staffers’ interactions with AI chatbots, and they demanded on Wednesday that members of the administration preserve their online records.
Democrats don’t currently have subpoena power to compel the government to release the records, but the request is an indication that the line of inquiry will be a priority for Democratic members of the House Oversight and Judiciary Committees should the party retake control of the chamber following next week’s midterm elections.
In a letter sent to the president by Reps. Jamie Raskin and Robert Garcia Wednesday, the congressmen — the top Democrats on Judiciary and Oversight, respectively — asked the president if the executive branch has been soliciting advice from Grok, the AI chatbot created by Elon Musk’s social media company X, to carry out official responsibilities.
“If the American President is relying on the policy advice of a Musk-designed, Hitler-friendly chatbot with a predilection for child pornography in order to carry out his official duties, Americans have a right to know,” the lawmakers wrote, referring to recent reports that claim Trump has been using the notoriously mercurial technology.
Trending
They added that their ultimate goal is to find out whether the Trump administration is leaking national security information to AI chatbots and determine what role AI may have in forming federal policy.
The congressmen are requesting that Trump, his staff and advisers archive prompts submitted to Grok, ChatGPT, Claude, Gemini, and Perplexity and the AI-generated responses to those prompts for posterity.
The lawmakers raised concerns about national security risks that could stem from AI chatbots being used throughout the federal government.
“Preserving the President’s and his advisors’ AI prompts is thus essential to preserving an accurate historical record and informing future administrations against the gaping national security vulnerabilities created by this Administration’s apparent reliance on unsecured, privately operated AI systems,” they wrote.
In a statement to The Washington Sun, the White House said it preserves all required records in accordance with the law.
Earlier this month, TechCrunch reported that Trump told Musk last year that he was spending hours talking to Grok.
According to the outlet, the president sought the chatbot’s advice on carrying out the capture of Venezuelan President Nicolás Maduro.
Grok apparently encouraged Trump to carry out the attack, stating Maduro is a “deeply unpopular dictator and that many Venezuelans would likely celebrate his downfall,” TechCrunch reported.
Trump later ordered Maduro’s capture without congressional approval.
Following the military action, Trump thought Grok was “ingenious,” according to the technology publication.
The lawmakers are concerned the president “consulted Grok—but not Congress—on your military deployment to Venezuela, circumventing Congress’s constitutional power to declare war.”
The lawmakers pointed to the U.S. military’s strike on an Iranian elementary school that killed at least 123 children as an example of what could happen as a result of errors perpetuated by AI. The strike stemmed from the military’s “overreliance on artificial-intelligence technology,” the letter says.
In another instance earlier this year, the U.S. military nearly intercepted a Chinese vessel in the Middle East that was suspected of transporting components for a nuclear weapons program. The claim stemmed from a report, but military officials later determined the author of the report used an AI chatbot that misidentified the ship’s cargo.
“This was not an isolated or aberrational mistake,” the Democrats say in the letter.
The congressmen assert the executive branch’s AI chats must be preserved under the 1978 Presidential Records Act.
The law establishes public ownership of records created by presidents and their staff. The National Archives and Records Administration stores and preserves presidential and vice presidential records generated in the course of their official duties.
The lawmakers argue that AI prompts submitted by staff fall within the definition of presidential records, and they are separately requesting a closed-door briefing regarding the executive branch’s use of AI by Nov. 11.
Earlier this year, the National Archives and Records Administration released guidance on AI use for federal agencies and legal counsel.
The guidance states that agencies’ use of AI platforms does not, in and of itself, result in the creation of federal records — although there are several instances in which “subsequent use of AI-generated materials in agency business may result in the creation of federal records.”
Jason R. Baron, the former director of litigation at the National Archives and Records Administration, said the federal government is in uncharted territory when it comes to preserving AI records.
“If staffers are drafting prompts on a Gen AI platform and then using the responses in the course of government business, then they may well be considered presidential records or federal records,” he said.
But just because a staffer uses AI doesn’t mean all communications are considered presidential or federal records. Chatbots should not be used for sensitive or classified information, Baron said.
“I can also conceive of there being prompts that one asks that remain personal to the person who’s asking them,” he added.
Author
