MADD Says the “Kill Switch” Is a Hoax. The Real Story Is Worse.

The federal government doesn’t enforce drunk driving laws. So who actually flips the switch in your car — and who gets the data afterward?

Mothers Against Drunk Driving sent supporters an email this week defending Section 24220 of the 2021 Infrastructure Investment and Jobs Act — what the organization brands the “HALT Drunk Driving Law” — against what it calls a coordinated misinformation campaign. The pitch: critics are spreading a “hoax” about a “government-controlled kill switch” in your car, Congress has twice rejected attempts to undo the law, and lawmakers should hold the line.

The email is rhetorically tight. It is also incomplete in ways that should bother anyone who actually reads the statute.

Start with the name

MADD calls it the HALT Drunk Driving Law. The actual statutory title is “Advanced Impaired Driving Technology,” and the operative text requires NHTSA to mandate “advanced drunk and impaired driving prevention technology” in new passenger vehicles. The Federal Register record for the rulemaking makes the scope explicit: NHTSA reads “impairment” to include alcohol, drugs, distraction, and drowsiness, and the agency has stated outright that Congress did not intend to limit the effort to alcohol. NHTSA narrowed its initial proposed rule to alcohol only because the BAC threshold is the one impairment state with clear legal definitions and measurement standards. The underlying authority is broader. That is not a hoax. That is the statute.

The federalism problem nobody is debating

Here is where MADD’s framing really falls apart. The federal government does not enforce drunk driving laws. NHTSA is not a police force. Its authority under 49 U.S.C. § 30111 governs what manufacturers must build into vehicles at the point of sale. Once the car is on the road, jurisdiction passes to states and municipalities — the actual cops, prosecutors, and DMVs of America.

So if there is no government kill switch, who is actually doing the switching?

The answer is the automaker. The detection system runs on the manufacturer’s hardware, executes the manufacturer’s software, and routes data through the manufacturer’s cloud. When your car decides you are impaired and refuses to start — or limits operation, or pulls over, depending on what NHTSA’s final rule allows — that decision is made by a private corporation’s proprietary code. The federal government wrote the requirement. The automaker owns the switch.

That changes the entire legal posture. Critics calling this a “government kill switch” are wrong on the specifics, and MADD is right to push back on that framing. But the corporate kill switch is, in some ways, harder to challenge than a government one would be. A government search requires a warrant. A government determination triggers due process. A private device generating data on your driving behavior triggers neither — it just creates a timestamped record that a state prosecutor can subpoena later if you are charged with anything. State and local enforcement become the downstream beneficiaries of an evidentiary stream they did not have to build, did not have to fund, and did not have to constitutionally justify. They just receive what the car already generated.

This is the federalism inversion nobody on either side is debating publicly. The federal government mandates a private surveillance system. State and local police inherit the data. The driver gets none of the procedural protections that would apply if any government actor had run the same query directly.

The drug problem MADD doesn’t talk about

Roughly half of fatally injured drivers in toxicology screenings test positive for at least one drug other than alcohol. The federal apparatus for detecting drug impairment is built around Drug Recognition Experts — police officers trained in a 12-step evaluation protocol developed by the LAPD in the 1970s and standardized by NHTSA.

A 2025 peer-reviewed Maryland study published in Traffic Injury Prevention found DRE opinions agreed with blood toxicology results at an overall success probability of 84.2 percent, falling to 80.5 percent when two or more drugs were involved. That is the strongest pro-DRE data in the literature. Other research is harsher. Maryland’s Carroll County Circuit Court has held that DRE identification “is not generally accepted as valid and reliable in the relevant scientific community,” and at least six states have refused to admit DRE officers as expert witnesses — including California, the state where the program was invented.

In other words: the existing human-driven system for identifying drug impairment, run by professionally trained officers, gets it wrong often enough that multiple state courts will not let those officers testify. And we want a steering-wheel sensor to do the same job?

What the sensors can actually detect

The detection technology Section 24220 contemplates — interior cameras tracking eye movement, steering-wheel touch sensors, software analyzing erratic driving patterns — can identify behavioral correlates of impairment. It cannot identify impairment itself.

The car does not know whether you are high, exhausted, distracted by a screaming toddler, or just a mediocre driver. It knows you drifted across the lane line three times in two minutes. The intervention follows from the behavior, not the cause. A driver who failed to sleep the night before a flight gets the same outcome as a driver who took THC gummies an hour ago. Whether that is good policy is a debate worth having. Pretending the technology distinguishes between them is not honest.

The thirty-dollar workaround

There is a more uncomfortable engineering question that Section 24220 supporters have not answered publicly: what happens when the driver does not cooperate with the sensors?

The leading touch-based prototype is the Driver Alcohol Detection System for Safety program, a public-private partnership between NHTSA and the Automotive Coalition for Traffic Safety that has been in development since 2008. DADSS uses near-infrared tissue spectroscopy — shining infrared light through the skin of the fingertip or palm, typically embedded in the gear shift, the start/stop button, or the steering wheel, and reading the absorption spectrum that returns. The system measures blood alcohol under the skin’s surface. It cannot read through fabric, leather, nitrile, or any other glove material. The optical path is interrupted. The sensor either refuses to register a reading or registers the wrong one.

This is not a fringe scenario. Any driver in the northern half of the country wears gloves for roughly five months a year. Construction workers, delivery drivers, mechanics, riders transferring from a motorcycle, and anyone with a skin condition routinely keep gloves on at the wheel. The DADSS program has not published a performance specification that addresses glove use, because there is no good answer. Either the car refuses to start when it cannot get a skin reading — which locks out a substantial share of legitimate drivers — or it defaults to allowing operation when the reading is ambiguous, which means the system is trivially defeated.

The camera track has the same problem from a different direction. Driver-monitoring systems already deployed in production vehicles — GM Super Cruise, Ford BlueCruise, and similar platforms — use an interior-facing camera to track pupil position, gaze direction, blink rate, and head pose. These metrics correlate with intoxication, drowsiness, and distraction. Polarized or mirrored sunglasses defeat pupil tracking outright and significantly degrade gaze estimation. Some manufacturers have invested in infrared-capable cameras that can see through certain lens coatings, but performance varies widely across lens types, and no Federal Motor Vehicle Safety Standard mandates a particular sensor wavelength or accuracy threshold for sunglass penetration. A driver in wraparound polarized lenses is, for many of these systems, functionally invisible at the eye level.

Combine the two — gloves and sunglasses — and the entire passive-detection layer collapses to whatever erratic-driving inference the lane-keeping and steering-angle sensors can produce. Which is to say: the same behavioral inference any officer with a radar gun and a line of sight could make from outside the car. The whole rationale for embedding the sensors in the vehicle in the first place — that the car can detect impairment a trained officer cannot — disappears.

This points to a deeper structural problem with the law. Passive detection systems are fundamentally voluntary. The driver who knows the sensors exist can defeat them with thirty dollars of gear from a gas station. The drivers who cannot or will not defeat them are disproportionately going to be the people who were not impaired to begin with: the careful, the law-abiding, the people who would have driven safely without any sensor at all. The actual target population — the chronically impaired driver who has decided to get behind the wheel anyway — is exactly the population most likely to learn the workaround. This is the same critique that has dogged ignition interlock devices for years. They get circumvented by sober helpers, mouth-alcohol confounds, and physical workarounds. Section 24220 inherits all of those problems while adding new ones.

The chemistry-level detection work is happening in the private sector

Actual chemistry-level detection — the kind that could tell you a driver has THC, opioids, or benzodiazepines in their system rather than just inferring impairment from how they are holding the wheel — is being developed outside the Section 24220 framework.

Cannabix Technologies Inc. (CSE: BLO; OTC: BLOZF), based in Vancouver, has spent more than a decade developing a delta-9 THC breath collection unit and launched new Marijuana Breath Test product videos in December 2025 aimed at workplace safety and roadside screening. NIST researchers published the first detection of cannabis in breath from edibles in July 2025, and a federally funded study released in April 2026 produced a 3D-printed colorimetric prototype for portable THC detection.

Ombra Group Inc., a developer of chemical and narcotic detection equipment, is pursuing technology in the same broad category of post-consumption substance detection. A public demonstration video shows the company’s sensor identifying THC under field conditions — the kind of direct chemical confirmation that no in-vehicle camera or steering-wheel sensor is capable of producing.

None of this is in your car. None of it is anticipated by the Section 24220 rulemaking. The federal mandate is racing toward an alcohol-focused passive-sensor implementation window in 2026–2027, while the harder problem — drug impairment, which already drives a substantial share of impaired-driving fatalities and is growing faster than alcohol — gets handed back to fallible DREs and toxicology labs operating on multi-week turnaround times.

What MADD’s email is really doing

When MADD writes that “anti-drunk driving technology will prevent these tragedies before they occur,” the operative word is drunk. For the roughly half of impaired-driving deaths involving something other than alcohol, the law promises a sensor that watches the steering wheel and reports its observations to a private cloud, where the data sits until someone with a subpoena comes looking for it.

The “kill switch” framing on the opposition side is sloppy. MADD is correct that no government agency is remotely disabling vehicles. But the substantive concern — a federally mandated, automaker-operated detection and disablement system that creates evidentiary records useful to state prosecutors, with no statutory privacy protections in place — is real. Dismissing it as a hoax is rhetorical sleight of hand. The policy question deserves a better debate than either side is currently offering.

That is not a hoax. That is the design.

DISCLOSURE

Ombra Group Inc. is part of the Kubera Technology Holdings Corp portfolio. Dominick Bianco is Founder and CEO of Kubera Technology Holdings Corp and Editor-in-Chief of NexfinityNews.com. Cannabix Technologies Inc. is referenced as a publicly traded comparable in the same product category and has no affiliation with NexfinityNews, Kubera Technology Holdings Corp, or Ombra Group Inc.

Key Takeaways

• Section 24220 is titled “Advanced Impaired Driving Technology,” not “drunk driving.” NHTSA reads its statutory authority to cover alcohol, drugs, distraction, and drowsiness.
• Federal authority ends at the assembly line. Once a vehicle is sold, enforcement is a state and local function.
• The “kill switch” is owned and operated by the automaker. The car decides; the government receives the data after the fact via subpoena.
• Drug Recognition Experts have a roughly 80–84 percent accuracy rate in the strongest studies. Multiple state courts refuse to admit DRE testimony.
• In-vehicle sensors detect erratic driving, not chemistry. Drug breathalyzer technology that could close that gap is being built in the private sector, outside the Section 24220 framework.
• The passive sensor architecture is defeatable with gloves and sunglasses. The drivers most likely to circumvent it are the chronically impaired — exactly the population the law claims to target.

Related Coverage from NexfinityNews

For readers tracking how private surveillance infrastructure interacts with state and federal enforcement, NexfinityNews has published related investigations:

• Big Brother on Every Corner: How Flock Safety Built America’s Largest Covert Surveillance Network — a 100,000-camera private ALPR network scanning 20 billion plates a month, feeding data to police, ICE, and federal agencies while resisting FOIA requests from the public it surveils.
• Chatrie v. United States: The Supreme Court Weighs the Fourth Amendment in the Age of Geofence Warrants — the Court’s first major digital-surveillance ruling since Carpenter, testing whether police can compel platforms to identify every device in a defined area during a defined time window.
• Trump Order Clears Path for U.S. Ibogaine Trials — the federal pivot toward expedited FDA review of Schedule I compounds for serious mental illness, including the regulatory tension between DEA scheduling and emerging therapeutic evidence.

Sources

1. Infrastructure Investment and Jobs Act of 2021, Section 24220 (Public Law 117–58). “Advanced Impaired Driving Technology.”
2. National Highway Traffic Safety Administration. Advance Notice of Proposed Rulemaking, “Advanced Impaired Driving Prevention Technology,” Federal Register, January 5, 2024.
3. NHTSA, Report to Congress: Advanced Impaired Driving Prevention Technology, February 2026.
4. Bhagat K., Kufera J., Auman K., Kaushik K., Vesselinov R. “An analysis of drug recognition expert evaluations and comparisons with police issued citations in Maryland, 2017–2021.” Traffic Injury Prevention, 2025.
5. Forensic Resources, North Carolina Office of Indigent Defense Services. “Drug Recognition Experts” resource page, updated 2025.
6. Cannabix Technologies Inc. “Cannabix Technologies Launches New Marijuana Breath Test Product Videos.” Press release, December 19, 2025.
7. National Institute of Standards and Technology. “NIST Makes First Detection of Cannabis in Breath From Edibles.” July 30, 2025.
8. Marijuana Moment. “Federally Funded Study Reveals Marijuana Breathalyzer Breakthrough With 3-D Printed Roadside Tool Able To Detect THC.” April 8, 2026.
9. Ombra Group THC Detection Demonstration. Public product demonstration video.
10. Driver Alcohol Detection System for Safety (DADSS) Program. Public-private partnership between NHTSA and the Automotive Coalition for Traffic Safety; technical documentation on tissue-spectroscopy touch sensors and breath-based ambient-air sensors.
11. H.R. 6704, Drunk Driving Prevention and Enforcement Act of 2025. 119th Congress.