The US Department of Transportation apparently thinks it’s a good idea to use artificial intelligence to draft rules impacting the safety of airplanes, cars, and pipelines, a ProPublica investigation revealed Monday.
It could be a problem if DOT becomes the first agency to use AI to draft rules, ProPublica pointed out, since AI is known to confidently get things wrong and hallucinate fabricated information. Staffers fear that any failure to catch AI errors could result in flawed laws, leading to lawsuits, injuries, or even deaths in the transportation system.
But the DOT’s top lawyer, Gregory Zerzan, isn’t worried about that, December meeting notes revealed, because the point isn’t for AI to be perfect. It’s for AI to help speed up the rule-making process, so that rules that take weeks or months to draft can instead be written within 30 days. According to Zerzan, DOT’s preferred tool, Google Gemini, can draft rules in under 30 minutes.
“We don’t need the perfect rule on XYZ,” Zerzan told DOT staffers at the meeting. “We don’t even need a very good rule on XYZ. We want good enough.”
DOT staffers “deeply skeptical” of Gemini
ProPublica spoke to experts and granted six DOT staffers anonymity to discuss their concerns about DOT’s use of Google Gemini to draft rules.
Some experts who monitor AI use in government told ProPublica that DOT could save time using Gemini as a research assistant “with plenty of supervision and transparency.” For example, at a presentation, DOT staffers were told that “most of what goes into the preambles of DOT regulatory documents is just ‘word salad,’” and “Gemini can do word salad.”
However, staffers told ProPublica they felt “deeply skeptical” that Gemini was up to the task. They emphasized that DOT rulemaking is “intricate work” requiring sometimes decades of “expertise in the subject at hand as well as in existing statutes, regulations, and case law.” Likely unsettling staffers further, ProPublica noted that a demonstration of Gemini’s rule-drafting produced a document missing key text, which a staffer would then have to fill in. Additionally, the DOT’s move comes after a year of AI hallucinations scrambling courts, with many lawyers fined and even judges admitting they can be fooled by fabricated information.
Any errors in the rules could have serious consequences. These rules “touch virtually every facet of transportation safety,” keeping “airplanes in the sky,” preventing “gas pipelines from exploding,” and stopping “freight trains carrying toxic chemicals from skidding off the rails,” ProPublica reported.
“It seems wildly irresponsible,” one staffer said.
Regardless of staffers’ concerns, DOT appears to be racing forward with the plan, ProPublica reported. The department has already used Gemini to draft a “still-unpublished Federal Aviation Administration rule, according to a DOT staffer briefed on the matter.”
Trump “very excited” about AI drafting rules
Donald Trump has urged federal agencies to adopt AI at a rapid pace, but nowhere in his orders has the president pushed for AI to draft laws, ProPublica noted.
However, Trump is “very excited” about the DOT initiative, Zerzan told staffers at the meeting, suggesting that Trump sees DOT as the “point of the spear” and expects other agencies will follow its lead.
At DOT, Trump likely hopes to see many rules quickly updated to modernize airways and roadways. In a report highlighting the Office of Science and Technology Policy’s biggest “wins” in 2025, the White House credited DOT with “replacing decades-old rules with flexible, innovation-friendly frameworks,” including fast-tracking rules to allow for more automated vehicles on the roads.
Right now, DOT expects that Gemini can be relied on to “handle 80 to 90 percent of the work of writing regulations,” ProPublica reported. Eventually all federal workers who rely on AI tools like Gemini to draft rules “would fall back into merely an oversight role, monitoring ‘AI-to-AI interactions,’” ProPublica reported.
Google silent on AI drafting safety rules
Google did not respond to Ars’ request to comment on this use case for Gemini, which could spread across government under Trump’s direction.
Instead, the tech giant posted a blog Monday, pitching Gemini for government more broadly, promising federal workers that AI would help with “creative problem-solving to the most critical aspects of their work.”
Google has been competing with AI rivals for government contracts, undercutting OpenAI and Anthropic’s $1 deals by offering a year of access to Gemini for $0.47.
The DOT contract seems important to Google. In a December blog, the company celebrated that DOT was “the first cabinet-level agency to fully transition its workforce away from legacy providers to Google Workspace with Gemini.”
At that time, Google suggested this move would help DOT “ensure the United States has the safest, most efficient, and modern transportation system in the world.”
Immediately, Google encouraged other federal leaders to launch their own efforts using Gemini.
“We are committed to supporting the DOT’s digital transformation and stand ready to help other federal leaders across the government adopt this blueprint for their own mission successes,” Google’s blog said.
DOT did not immediately respond to Ars’ request for comment.
Social network UpScrolled sees surge in downloads following TikTok’s US takeover