The U.S. Department of Transportation (DOT) is planning to harness artificial intelligence to draft federal transportation regulations, as revealed in internal agency records and interviews with staff. The initiative was introduced in December during a demonstration highlighting AI’s potential to transform regulatory drafting processes. Daniel Cohen, an agency attorney, conveyed enthusiasm for the technology’s ability to enhance efficiency, stating it could help staff “do our job better and faster.”
Discussions around the initiative continued among DOT leadership last week, with general counsel Gregory Zerzan noting President Donald Trump is “very excited” about the prospect. Zerzan described the DOT as “the point of the spear” in a broader federal initiative to integrate AI, asserting that the department would be the first agency to fully leverage AI for rule drafting. He emphasized not the quality of the regulations but the quantity, declaring, “We want good enough,” and “We’re flooding the zone.”
The proposal has raised concerns among some DOT employees. The agency’s regulations impact transportation safety across numerous sectors, such as aviation, gas pipelines, and hazardous freight. Employees have questioned the wisdom of delegating such critical responsibilities to an emerging technology known for its inaccuracies. Proponents, however, argue that AI’s speed could streamline the typically lengthy process of drafting complex regulations, which can take months or even years. They claim that DOT’s use of Google Gemini could allow for the generation of proposed rules in mere minutes.
Zerzan reiterated ambitions to significantly shorten the timeframe for creating transportation regulations, aiming for a draft ready for review within 30 days. He indicated that the expectation is for staff to produce a draft rule from Gemini in no more than 20 minutes. This push for rapid regulatory drafting aligns with the Trump administration’s broader efforts to incorporate AI into federal workflows, which has seen previous administrations use such technology for tasks like document translation and data analysis.
While the current administration’s enthusiasm for AI is notable, it is not unprecedented. Trump issued several executive orders last year supporting AI initiatives. In April, Office of Management and Budget Director Russell Vought circulated a memo urging the acceleration of AI’s use in government. However, the documents did not specifically advocate for employing AI to draft regulations, as the DOT plans to do.
Concrete actions have already been taken; DOT has reportedly used AI to draft an unpublished Federal Aviation Administration rule. Despite this, skeptics argue that large language models like Gemini and ChatGPT are ill-suited for the intricate tasks of governance, which often require nuanced human reasoning. Proponents see AI as a means to automate mundane tasks and improve efficiencies within a sluggish bureaucratic framework.
At a recent AI summit in Northern Virginia, federal technology officials discussed fostering an “AI culture” within government and enhancing workforce capabilities to utilize the technology. Justin Ubert, from DOT’s Federal Transit Administration, highlighted the potential for AI to eliminate human bottlenecks, predicting a future where human roles may primarily involve monitoring AI interactions.
The December demonstration, attended by over 100 DOT employees, generated mixed reactions. During the session, the presenter claimed that Gemini could handle 80% to 90% of regulatory writing tasks, allowing staff to focus on the remaining elements. However, attendees noted that the AI-generated draft was missing essential text for the Code of Federal Regulations. Despite concerns about the potential for “hallucinations”—inaccurate or nonsensical outputs from AI—the presenter suggested that DOT staff would simply proofread the AI’s outputs.
Some employees voiced skepticism about the entire approach, stressing that effective rulemaking necessitates deep subject matter knowledge and an understanding of existing regulations. Errors in transportation regulations could result in severe consequences. Mike Horton, the former acting chief AI officer at DOT, likened the plan to relying on a “high school intern” for rulemaking and cautioned that a hasty approach could jeopardize public safety.
Experts in AI application within government hold varied opinions on the DOT’s strategy. While some believe that AI can serve as a helpful research assistant with appropriate oversight, others warn against excessive reliance on technology, as it may undermine the quality of critical regulations. Bridget Dooling, a professor at Ohio State University, noted, “Just because these tools can produce a lot of words doesn’t mean that those words add up to a high-quality government decision.”
With federal workforce reductions leading to a loss of subject-matter experts, critics argue that the plan could be especially detrimental. DOT has seen a decrease of nearly 4,000 employees since Trump’s return to office, which includes over 100 attorneys. The initiative also aligns with broader calls for increased government efficiency, notably from Elon Musk’s Department of Government Efficiency, which previously advocated for using AI to automate regulation drafting.
As the White House has not clarified whether similar AI initiatives will be adopted by other federal agencies, skepticism persists among technology officials regarding the DOT’s claims of being a leader in AI integration. Concerns about the implications of such rapid technological adoption in governance continue to grow, raising fundamental questions about the intersection of technology and public safety.
See also
OpenAI’s Rogue AI Safeguards: Decoding the 2025 Safety Revolution
US AI Developments in 2025 Set Stage for 2026 Compliance Challenges and Strategies
Trump Drafts Executive Order to Block State AI Regulations, Centralizing Authority Under Federal Control
California Court Rules AI Misuse Heightens Lawyer’s Responsibilities in Noland Case
Policymakers Urged to Establish Comprehensive Regulations for AI in Mental Health
