DOGE Workers Accused of Dumping Sensitive Data into AI, Asking Software to Slash Education Department Budget

Members of Elon Musk's Department of Government Efficiency (DOGE) reportedly fed sensitive government data into artificial intelligence tools, relying on AI to recommend drastic budget cuts—particularly targeting the Department of Education.

According to multiple U.S. officials, DOGE agents, mostly in their twenties with little prior government experience, have gained administrative access to highly restricted personnel records at the Office of Personnel Management (OPM). Now, internal concerns are growing that these same operatives have been running confidential government data through AI models, risking external access to sensitive data.

"The idea that unvetted individuals with full access to federal employee records are inputting them into AI systems is beyond reckless," a former intelligence official told the Washington Post. "It's a nightmare scenario for data security and government oversight."

Sources familiar with the matter who asked to remain anonymous say that in at least one instance, DOGE team members uploaded agency staffing and expenditure reports into an AI platform and asked it to identify "inefficiencies," as reported by the Post.

The AI reportedly suggested slashing the Department of Education's budget by over 50%, prompting DOGE operatives to begin drafting proposals to gut the agency.

The revelations have further fueled concerns about DOGE's unchecked authority over federal agencies. "This isn't governance—it's a tech-bro experiment on a national scale," one senior OPM official told the outlet.

The White House has not commented on whether DOGE agents have been authorized to use AI for budget planning or data analysis.

Earlier this week, the US Treasury insisted DOGE had been given "read-only" access to the data. However, the Post obtained records they say show some members of DOGE had "administrative" access to OPM computer systems, which would enable them to install and modify software or alter internal documentation of their own activities.

Meanwhile, cybersecurity experts warn that if sensitive personnel files were uploaded to AI models, they may now be accessible to third parties, including foreign adversaries.