Congress required an impaired-driving detection standard in new vehicles by November 2024 — as of August 2026 no rule has been proposed

What is happening?

The Infrastructure Investment and Jobs Act, enacted 2021-11-15, directs the Secretary of Transportation to issue a final Federal Motor Vehicle Safety Standard requiring advanced drunk and impaired driving prevention technology in new passenger vehicles. Section 24220 of that act sets the deadline at three years from enactment, which fell on 2024-11-15. No such standard has been issued.

What happened instead is a short record. NHTSA published an advance notice of proposed rulemaking on 2024-01-05 at 89 FR 830, under docket NHTSA-2022-0079 and RIN 2127-AM50, and closed the comment period on 2024-03-05. Nothing further followed. The federal document index for this subject carries no notice of proposed rulemaking and no final rule at any date through 2026-07-28.

The regulatory agenda entry for that RIN still records the rulemaking stage as prerule. Its statutory-deadline field records the 2024-11-15 date, so the agency register carries the requirement and the miss in the same place. Its timetable projects no date for a proposed rule and no date for a final rule. The only forward milestone it names is analyzing the comments on the 2024 notice, dated October 2026. An earlier edition of the same entry dated that same milestone April 2025.

NTSB recommendation H-22-22, issued with a highway investigation report on 2022-08-22 after a crash on 2021-01-01 in Avenal, California in which nine people died including seven children, asked NHTSA to require the same class of technology. That report records that Congress had already legislated the requirement, and that the board joins Congress in pressing for progress in this area.

Whose problem is this?

The body that must write the standard is the same body that received a statutory command to write it, and the only party the statute names as a recipient of anything when the deadline passes is Congress itself.

Where does this problem end?

The boundary matters because the requirement is not a proposal from outside government but a duty that Congress placed on an executive agency with a date attached to it.

What is the state now, and what should it be?

Now

Should be — the statute states the target itself, so the standard being asked for does not have to be inferred. Section 24220 defines the technology as a system that can passively monitor the performance of a driver to identify impairment, or passively and accurately detect whether the blood alcohol concentration of the driver is at or above the legal limit of .08 g/dL, and that in either case can prevent or limit vehicle operation when impairment is detected. The statute also fixes when such a rule would bind: a compliance date not earlier than two years and not more than three years after the rule issues. A rule issued in 2026 would therefore not bind vehicles before 2028 or 2029, and then only vehicles built after that.

How big is it?

Federal counts for a single year, 2021:

The first two rows are two federal counts of the same year under two definitions, not a range and not a margin of error. The gap between them is definitional, and both are stated by the same notice.

One further figure sits below the threshold the statute names. In 2020, 2,041 people were killed in crashes where a driver measured between .01 and .07 g/dL. That band is outside what a system set to the .08 legal limit is built to catch, and it is a different year from the counts above, so it is not added to them.

The ceiling on what the technology could remove from those counts has been estimated. The safety board report cites work published in 2021 finding that systems keeping drivers at or above .08 off the road could prevent more than 9,000 deaths a year, and that direct measurement keeping any driver with measurable alcohol off the road could prevent nearly 12,000 a year. Both are full-fitment figures and neither is a count of anything that has happened.

Under what conditions does it arise?

1. The consequence the statute attaches to delay is a report. Section 24220 requires reports to the Senate Commerce committee and the House Energy and Commerce committee no later than the deadline and at least annually after that until the rule issues. It attaches nothing else. There is no penalty, no default standard that takes effect on its own, and no consequence that operates without a further decision by someone.

2. The statute also supplies a lawful way to be late. The Secretary may extend the deadline as necessary, up to a ceiling of three years past it, which places the outer limit at 2027-11-15. Two federal documents describe the reporting duty differently. The statutory text ties it to the date the rule issues inside that window; the advance notice describes the duty as running until a final rule issues or ten years have expired from enactment, which would reach into 2031. The statutory text is the operative one, and the difference between the two descriptions is itself part of the record.

3. The agency states technical reasons. The advance notice says drugged driving is not being considered there because of technology immaturity and a lack of testing protocols, and that challenges remain in distinguishing between different impairment states, avoiding false positives, and determining appropriate prevention countermeasures. It points to the Driver Alcohol Detection System for Safety research program as the vehicle for resolving them.

4. The record is unusually large. The docket had received 18,367 comments as of 2026-04-25, against a comment window that closed on 2024-03-05. Reading a record of that size is the milestone the agenda is still projecting forward, and it is the milestone that moved between editions.

5. The lag is built into the instrument. Even after a rule issues, the statute sets compliance no earlier than two years later. The safety board report cites an estimate that even with the technology required as standard equipment in all new vehicles it would take twelve years to reach even half of the lifesaving potential, with 1,000 to 1,300 lives a year saved within three years of a mandate, about 2,000 to 2,600 within six years, and about 4,600 to 5,900 within twelve.

What has been tried?

Two of these are commands and the rest are steps inside a process that has not yet produced the document the commands asked for.

What was found?

Why is it still unsolved?

Enforcement absent — the command exists, the date for obeying it has passed, and the only thing the statute attaches to that date passing is a report.

This is not a case where nobody has decided what should happen. Congress decided in 2021 and wrote the decision into law with a date on it. The agency that has to carry it out recorded the same date in the statutory-deadline field of its own regulatory agenda entry, both in the edition published before the date and in the edition published after. Two federal documents agree on what was required and when. What is missing is the document that would make it operative.

The first part of the gap is that the statute names no consequence that operates on its own. Reports to two congressional committees are the whole of it. A report goes to bodies that already set the deadline, and nothing in the text turns the absence of a rule into a rule, a penalty, or a date certain.

The second part is that the statute supplies a lawful route to lateness and then runs out. The Secretary may extend, up to 2027-11-15. Beyond that ceiling the statute is silent about what follows.

The third part is that the process emits no signal a reader can tell apart from work in progress. The stage recorded in the agenda is prerule, the same stage recorded before the deadline. The forward milestone is reading comments, and that milestone moved eighteen months between two editions of the same entry. The most recent activity in the federal record is a study that the notice announcing it describes as not immediately intended to inform regulations or policy. Each of those is an ordinary thing for an agency to be doing, and together they are indistinguishable from an agency that has stopped.

The fourth part is the one that shapes every rulemaking of this kind. The parties who would pay can price the equipment to the dollar. The people who would not die in the years after a compliance date cannot be named in advance, so their side of the ledger arrives as a modeled estimate while the other side arrives as an invoice.

What observation would mean it is solved?

Candidates — (a) NHTSA publishes a final rule in the Federal Register with a stated compliance date (b) the safety board reclassifies H-22-22 to closed with acceptable action (c) the share of vehicles on the road carrying the technology rises.

(a) alone is not enough, and the statute itself says why. The compliance date it permits is not earlier than two years and not more than three years after the rule issues. A rule issued in 2026 would therefore bind no vehicle before 2028 or 2029, and then only vehicles built after that point.

(c) is where the effect actually lives, and it is slow by arithmetic rather than by choice. The safety board report cites an estimate that even with the technology required as standard equipment in all new vehicles it would take twelve years to reach even half of the lifesaving potential, with 1,000 to 1,300 lives a year saved within three years of a mandate, about 2,000 to 2,600 within six years, and about 4,600 to 5,900 within twelve. Fleet turnover sets that pace, and no rule can change it.

(b) alone counts a label. The classification records what one federal board thinks of the response of another agency. It is a useful signal and it is not a vehicle standard.

The three have to be read together, and (a) has to be read with the compliance date attached to it rather than with the date it is signed.

What is it connected to?

Fills with researchdriver monitoring and in-vehicle data privacy · the per se blood alcohol limit as a separate highway safety axis · drugged driving detection, which the advance notice sets aside · and statutory deadlines on agency rulemaking generally. Relation type and evidence grade are not established by any source reached here.

What these sources do not say

• Whether the reports to Congress were filed. Section 24220 makes reports to the Senate Commerce committee and the House Energy and Commerce committee the consequence of the deadline passing, and no source reached here shows whether any such report exists. The departmental index of reports to Congress returns HTTP 403 to automated retrieval, the agency site returns 403 across paths, and the congressional site blocks automated retrieval. Whether the one consequence the statute names actually operated is therefore not established either way.
• Whether the Secretary formally invoked the extension. The same access refusals apply, and no notice announcing an extension appears in the federal document index reached.
• When H-22-22 was reclassified and on what stated ground. The per-recommendation lookup returns an internal error for every identifier submitted, including identifiers that certainly exist, and the replacement database renders as an empty page. A board release dated 2023-12-11 confirms the classification was in force by that date and states neither the date of the change nor a ground for it.
• The wider recommendation history. The report that issued H-22-22 classifies H-17-24 and H-12-33 as open with an unacceptable response, and H-12-43 and H-12-48 as closed with unacceptable action, the last two on the ground that in the ten years since they were first issued neither passive breath-based nor touch-based detection had reached sufficient development for vehicle implementation or field testing. No fuller history is retrievable while the recommendation database is down.
• A false-positive rate. The advance notice names avoiding false positives as an open challenge and publishes no rate, no target and no test protocol. The performance specifications of the research program it points to were not reachable.
• The added cost per vehicle. The advance notice carries societal crash costs and no equipment cost. That gap is structural rather than an oversight, because the cost analysis for a rule is produced at the proposal stage and this rulemaking has not reached it.
• How often an unacceptable-response classification ends in implementation. The board publishes recommendation statistics, but the pages that would carry them return 404 and 401 and the searchable database is down.
• An official definition of the classification itself. The board pages that define recommendation status classifications return 404 and 401, so the meaning of the label rests on no source reached here.
• Legislative, audit and oversight developments in 2025 and 2026. Congressional, audit and inspector-general sites block automated retrieval, so nothing can be recorded here either way about hearings, reviews, or bills to repeal, extend or fund the mandate.

See the evidence

Read directly and end to end — the statutory note carrying section 24220, the full text of the advance notice, the highway investigation report that issued H-22-22, and both editions of the regulatory agenda entry for the rulemaking. Those are primary documents, and the evidence_tier: secondary declaration reflects the document as a whole rather than each row: the fate of the reports owed to Congress and the definition of the status label are beyond reach because the hosts holding them refused automated retrieval, and the classification history of the recommendation is beyond reach for a different reason, that the recommendation database returns an internal error to every identifier submitted. The statute and the advance notice agree on the deadline, and the agenda entries carry that same date in their statutory-deadline field, so the central facts of this document rest on three independent federal records rather than on any account of them. Where the sources diverge the divergence is left visible rather than resolved: the statute caps the extension at three years past the deadline while the advance notice describes the reporting duty as running until a final rule issues or ten years have expired from enactment, and the two 2021 fatality counts of 13,384 and approximately 12,581 rest on different definitions that the notice does not reconcile. The largest unresolved item is not a divergence but a hole: the statute names reports to Congress as the consequence of the deadline passing, and the sites that would show whether those reports exist refuse automated retrieval. This is a Path A output (research-based definition), so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range 12,581–13,384 As of 2021

Sensitivity The interval is not a confidence interval and not a margin of error. It is the distance between two federal definitions of the same calendar year published in the same document, one counting deaths attributed to alcohol impairment and the other counting all deaths in crashes involving a driver at or above the .08 threshold, and that document does not reconcile them. Four limits run in different directions. First, this counts deaths only; injuries in the same crashes are not counted anywhere in the sources reached, so the true affected population is larger by an unknown amount. Second, a single year understates the standing exposure: the annual average across 2011 through 2020 was almost 10,500, so 2021 sits above the preceding decade. Third, in the other direction, this is not the number the technology would remove. The modelled ceiling at full fitment is more than 9,000 deaths a year for systems set at the .08 threshold and nearly 12,000 a year for direct measurement keeping any driver with measurable alcohol off the road, and the realised figure ramps far more slowly, at 1,000 to 1,300 lives a year within three years of a mandate and about 4,600 to 5,900 within twelve years, because fleet turnover rather than the rule sets the pace. Fourth, a further 2,041 people were killed in 2020 in crashes where a driver measured between .01 and .07 g/dL, a band below the threshold the statute names; that figure is a different year and a different definition and is deliberately not added here.

Regional breakdown No source reached here gives a state-by-state or region-by-region split of alcohol-involved traffic fatalities. Only national totals were published in the documents opened. Dividing the national figure by state population would be proportional allocation, and it would be wrong in both directions at once, because alcohol-involved fatality rates and vehicle miles travelled per person vary widely between states and neither varies with population.