Dems fret over DOGE feeding sensitive data into random AI

A group of 48 House Democrats is concerned that Elon Musk's cost-trimmers at DOGE are being careless in their use of AI to help figure out where to slash, creating security risks and giving the oligarch's artificial intelligence lab an inside track to train its models on government info.

Led by Representatives Don Beyer (D-VA), Mike Levin (D-CA), and Melanie Stansbury (D-NM), the Dems wrote a stern letter to the White House's Office of Management and Budget (OMB), saying that DOGE's reported use of AI runs afoul of several federal laws and the OMB's own AI directives, and is unlikely to be in compliance with FedRAMP standards for cloud software security.

Beyond that, the Reps believe Musk, who as well as SpaceX and Tesla runs OpenAI rival xAI, is self-dealing by using his own Grok-2 AI model to ingest government data.

"It is clear that DOGE's use of AI clearly does not meet the standards the previous memoranda set," the Dems said. "We ask that you immediately terminate any use of AI systems that have not been approved ... or that do not comply with existing laws."

The letter lists a number of instances where DOGE, in the Democrats' view, played fast and loose with AI in government agencies.

One such complaint involves a White House-based DOGE aide, doing double duty as a SpaceX employee, who reportedly created a chat-based AI assistant powered by xAI's Grok-2 - and hosted it on a subdomain of his personal website. The bot was programmed to "assist government personnel in identifying waste and improving efficiency."

In another instance, the Dems aren't happy about a report that a DOGE AI chatbot built on Anthropic and Meta models was used to analyze government contract data at the US General Services Administration, ostensibly giving government contractor Elon Musk access to competition insights.

More generally, the Democrats are worried about the sheer privacy and security risk of feeding sensitive government data, potentially including personally identifiable information about government employees and financial data, into uncontrolled and unvetted external AI systems.

"Without proper protections, feeding sensitive data into an AI system puts it into the possession of a system's operator - a massive breach of public and employee trust and an increase in cybersecurity risks surrounding that data," the letter reads.

That said, the memo's signatories don't want to come across as total AI alarmists.

"While we support the federal government integrating new, approved AI technologies that can improve efficiency or efficacy, we cannot sacrifice security, privacy, and appropriate use standards when interacting with federal data," they argue in the letter.

In short, feel free to use AI to streamline the work process - but for pete's sake, do it properly. The letter closes with a list of blunt questions for the Republican regime, including whether DOGE is using AI, and if so, which models and how.

This isn't the first letter Dems have sent on the subject. Representative Gerald Connolly (D-VA) fired off letters to two dozen federal agencies in March asking them to explain whether DOGE was feeding government data into unapproved AI models. Connolly, who's also a signatory on this latest letter, cited the exact same trio of laws in his letter that Beyer and company suspected DOGE of breaking - the Privacy Act of 1974, the Federal Information Security Management Act, and the E-Government Act of 2002 - and was concerned about lack of FedRAMP approval as well.

A spokesperson for Rep Beyer's office told The Register that, if the letter doesn't result in an immediate end to unauthorized DOGE AI deployments, "Congressman Beyer may consider further action."

DOGE - which is the Trump-blessed unit of operatives scouring the federal government for spending cuts to recommend - has faced a number of legal challenges, and has at least once been forced to slow down and do things by the book. Earlier this week, a judge partly lifted a federal injunction and allowed one DOGE staffer to access sensitive data at the US Treasury after he went through the training and vetting required for federal employees.

OMB didn't respond to questions for this story. ®