Congress Ordered Drunk-Driving Tech in Every New Car by 2024. NHTSA Has Produced 18,367 Comments and Zero Rules.

Section 24220 of the infrastructure law told NHTSA to write a federal safety standard requiring passive drunk-driving detection in every new passenger vehicle, with a final rule due November 15, 2024. That deadline is now 661 days in the past, and the rulemaking docket contains exactly zero final rules.^[1] I ran the numbers on what happened in those 661 days, then ran them again, and they didn't get better.

NHTSA counted 11,904 alcohol-impaired driving deaths in 2024, one every 44 minutes.^[2] Spread across a leap year, that is 32.5 deaths a day, and 661 days at that rate gives roughly 21,500 people dead in exactly the kind of crash this law was written to prevent. Numbers like that make you want to point fingers, so hold that instinct for one paragraph.

My arithmetic hides a wrinkle: even if NHTSA had issued the rule on the day Congress demanded, not one car on the road today would carry the system, because the statute gives automakers roughly three years after the final rule to comply. Deadline math, done honestly, puts the first compliant model years around late 2027. None of those 21,500 deaths were caused by the delay; it shifted the horizon. That distinction matters, because the rest of this article is about what the shift costs.

Start with what NHTSA has actually done. In the docket, NHTSA-2022-0079, the status reads Pending, last touched June 9, 2026.^[3] An advance notice went out in January 2024 and collected 18,367 public comments, one of the most-commented dockets of the decade, followed by a December 2025 Paperwork Reduction Act notice about designing a dashboard telltale for the impairment system. Read that again, because it means the most recent substantive activity on a mandate for drunk-driving detection technology is paperwork for the warning light on a device that does not commercially exist.

That last claim is not mine; it belongs to NHTSA. According to the agency's February 2026 report to Congress, no commercially available technology detects driver alcohol impairment accurately and passively, and current systems have yet to demonstrate the precision, speed, and reliability the mandate requires.^[4] False positives around the legal limit are the sticking point, and the error rate there, in the agency's own words, would be unacceptably high.^[5] A car that strands sober drivers at the grocery store is not a safety device; it is a lawsuit factory with wheels.

This is also where the program's own promises went to die. DADSS, the government-industry sensor partnership, once projected its ambient-air system as a factory option as early as 2025, and that year came and went.^[6] Real-world testing amounts to a handful of pilot vehicles in Virginia and Maryland plus a Schneider trucking trial, logging breath samples and operating hours while the commercial timeline slides. Sensors work in demos but not, passively and reliably, in your driveway, and those are different planets.

Meanwhile the law is very much alive. A House amendment to defund the mandate failed 164 to 268 in January, a separate repeal bill sits in committee, and another amendment rides along on a surveillance bill.^[7] Every vote leaves the mandate standing, which makes its current state a special category: legally alive, technically unborn. Congress ordered a technology into existence on a schedule, and the technology declined to cooperate.

Now the counterargument, at full strength, because the numbers deserve it. Arguably the delay is the responsible move: under the Motor Vehicle Safety Act, standards must be practicable, and mandating a sensor that misfires near 0.08 would strand sober drivers, breed public contempt, and hand automakers a compliance nightmare. That oft-cited IIHS figure of 9,409 lives a year from perfect alcohol detection assumes the fantasy scenario of flawless technology in 100 percent of vehicles instantly, while real fleet turnover is slow, so the honest first-decade yield is a fraction of the headline number.^[8] Congress anticipated exactly this and wrote an extension mechanism into the statute, which makes calling it a missed deadline legally incomplete: the deadline was aspirational, and the extension is the law working as written.

What actually moves the needle while the docket idles comes down to two things. First, the proven tool exists and needs no sensor breakthrough: ignition interlocks for convicted offenders, which IIHS estimates could avert up to 837 deaths a year on their own, a number that requires zero technological miracles.^[8] If your state doesn't require them for first offenders, that is the law to change. Second, when NHTSA finally opens the notice of proposed rulemaking, it will be the highest-leverage public comment period in vehicle safety this decade, and 18,367 comments already arrived on the advance notice. Make the proposal beat it.

What this article did not prove: the 21,500 figure projects 2024 fatality rates forward and assumes nothing changed, since 2025 and 2026 final data are not yet published. Note that the IIHS 9,409 estimate rests on 2015 to 2018 crash data and a perfect-tech fantasy, not a forecast. No specific death can be attributed to the delay, and as argued above, the counterfactual fails on the compliance timeline. DADSS progress figures come from program announcements, not independent audits. My headline number is damning, my counterargument is better, and both are in the article, which is how it should be.