At a Glance
- Trump administration plans to let AI draft federal transportation regulations.
- The plan could cut rule-making time from months to 30 days.
- Staff and experts warn that speed may come at the cost of safety and quality.
Why it matters: The rules set by the Department of Transportation (DOT) govern air, rail, and road safety. If AI tools produce inaccurate or incomplete rules, the consequences could affect millions of travelers.
The Trump administration is preparing to let artificial intelligence write federal transportation regulations, according to Department of Transportation records and interviews with six agency staffers. The proposal, first shown to DOT employees in a December demonstration, promises to speed rule-making from months to a few days. While the idea appeals to those who value efficiency, critics worry that the technology may produce errors that could jeopardize safety.
Background
The DOT has long used AI for tasks such as translating documents and analyzing public comments. The current administration, however, has intensified its push for AI, issuing executive orders and an “AI Action Plan” that called for broader adoption across federal agencies. The new plan marks the first time the DOT intends to use AI to draft entire regulations.
In a meeting last week, the agency’s general counsel, Gregory Zerzan, told colleagues that President Donald Trump was “very excited about this initiative.” He described the DOT as the “point of the spear,” the first agency fully enabled to use AI for rulemaking. Zerzan emphasized the quantity of regulations AI could produce, saying the agency “doesn’t need the perfect rule on XYZ… we want good enough.”
How the Plan Would Work
During the December presentation, more than 100 DOT employees-division heads, attorneys, and rule-making staff-were shown how Google Gemini could draft a Notice of Proposed Rulemaking in seconds. The presenter claimed Gemini could handle 80% to 90% of the writing, leaving humans to proofread and add detail.
The demonstration included a live example: an audience member suggested a topic, the presenter fed keywords into Gemini, and the AI produced a draft document. The draft lacked the precise text that goes into the Code of Federal Regulations, a point that raised concerns among attendees.
The goal is to compress the timeline for producing a complete draft ready for review by the Office of Information and Regulatory Affairs to just 30 days. Zerzan said the process should take no more than 20 minutes to generate a draft rule.
Staff Reactions
Several DOT staffers expressed skepticism. One anonymous rule writer said the presentation was “wildly irresponsible.” Mike Horton, the agency’s former acting chief AI officer, compared the plan to “having a high school intern that’s doing your rulemaking.” He warned that the agency’s leaders “want to go fast and break things, but going fast and breaking things means people are going to get hurt.”
Staffers highlighted that transportation rules affect air safety, gas pipelines, and freight trains carrying toxic chemicals. A rule that contains hallucinations-false statements produced by large language models-could lead to lawsuits or accidents.
Expert Opinions
Academics and researchers offered mixed views. Bridget Dooling, a professor at Ohio State University, said that while AI can be a useful research assistant, “just because these tools can produce a lot of words doesn’t mean that those words add up to a high-quality government decision.” She urged careful supervision and transparency.
Ben Winters, the AI and privacy director at the Consumer Federation of America, pointed out that the DOT has lost nearly $4,000 of its 57,000 employees since Trump returned to the White House, including more than 100 attorneys. He argued that the agency’s expertise is already thin, making reliance on AI risky.
Administrative Context
The DOT’s plan aligns with the administration’s broader AI agenda. In April, the Office of Management and Budget circulated a memo urging faster AI adoption. Three months later, an “AI Action Plan” reiterated this push. None of these documents, however, explicitly called for AI to draft regulations.
The DOT has already used AI to draft an unpublished Federal Aviation Administration rule, according to a DOT staffer. The agency’s acting chief AI officer, Brian Brotsos, declined to comment, directing questions to the DOT press office.
The White House has not confirmed whether other agencies will adopt similar AI-driven rulemaking. Four top technology officials said they were unaware of any such plans, and two expressed skepticism about the DOT’s “point of the spear” claim.
Key Takeaways
- The Trump administration plans to let AI draft federal transportation regulations.
- The initiative could reduce rule-making time from months to 30 days.
- Staff and experts warn that speed may compromise safety and rule quality.
- The DOT’s move follows a broader push for AI across federal agencies, but no other agency has announced a similar plan.
- The agency’s workforce has shrunk by nearly $4,000 employees, raising concerns about expertise and oversight.
The DOT’s proposal sits at the intersection of technology and public safety. While AI offers speed, the stakes of transportation regulation demand careful scrutiny and rigorous human oversight.
Timeline of Events
| Date | Event |
|---|---|
| December | DOT demonstration of Google Gemini to staff |
| Last week | Meeting with agency leadership, Zerzan’s remarks |
| April | OMB memo urging faster AI adoption |
| July | Washington Post reports on DOGE presentation (unrelated to DOT) |
The plan’s future will hinge on how the DOT balances efficiency with the rigorous standards required for transportation safety.
Closing
The Trump administration’s AI rule-making proposal reflects a broader trend toward automation in federal agencies. Whether the technology can deliver accurate, reliable regulations without compromising safety remains to be seen.
References
- DOT records and meeting notes reviewed by Julia N. Fairmont.
- Interviews with six agency staffers.
- Statements from Gregory Zerzan, Daniel Cohen, and others.
- Comments from academics and industry experts.
