Four People Oversee Whether 6 Million Teslas Can Drive Themselves. Congress Wants to Cut Their Budget.
Before February 2025, seven people at the National Highway Traffic Safety Administration worked full-time on investigating whether autonomous and semi-autonomous vehicles were safe enough for American roads.[1] Seven people for every Tesla on Autopilot, every Waymo robotaxi, every GM Cruise pod, every "Full Self-Driving" beta tester merging onto a highway while a camera system decided whether the sun was too bright to see. After the federal workforce cuts, that team dropped to four.
Four people. The vehicles they are responsible for investigating did not shrink by 43% to match. Those vehicles grew.
What Four People Are Supposed to Investigate
NHTSA currently has three active investigations touching Tesla's driver-assistance systems, covering a combined 6.08 million vehicles:[2]
EA26002 is the big one. An engineering analysis, which is NHTSA's final investigative stage before the agency can order a mandatory recall, covering approximately 3.2 million Tesla vehicles equipped with Full Self-Driving. Opened after four reported collisions in low-visibility conditions, including a fatal crash that killed Johna Story, a 71-year-old grandmother struck while directing traffic around another accident. NHTSA found Tesla's system "fails to detect and/or warn the driver appropriately under degraded visibility conditions such as glare and airborne obscurants."[2] Tesla settled the Story family's lawsuit this month. Terms undisclosed.[3]
PE25012 covers 2.88 million vehicles over more than fifty reports of FSD committing traffic violations: running red lights, crossing into oncoming lanes, behaviors that would get a human driver arrested.[2]
And then Katy, Texas happened. On June 19, a Model 3 plowed into a residential home at 73 miles per hour and killed Maria Avila, a 76-year-old woman, inside her own house.[4] Both NHTSA and the National Transportation Safety Board opened separate investigations into the same crash, a rare move that underscores how seriously federal regulators are taking FSD-related incidents. Two agencies. One crash. A dead woman who was not in a car, not on a road, and not a participant in the transportation system that killed her.
Ashok Elluswamy, Tesla's VP of AI software, posted on X that the driver "manually overrode self-driving by pressing the accelerator all the way to 100%."[4] The driver told the Harris County Sheriff's Office that an automated driving system was engaged.[4] Someone is wrong, and it falls to a four-person team to determine who.
The Math That Should Alarm You
Before the cuts, those seven investigators each covered roughly 870,000 vehicles under active AV investigation. After: 1.52 million vehicles per person. That ratio degraded by 75% while the caseload stayed constant or grew.
But the autonomous vehicle team does not operate in a vacuum. The entire agency lost 28% of its workforce, dropping from 772 employees to approximately 555, through what were described as "financial incentive programs" for voluntary departures rather than direct layoffs.[1] The total number of vehicles under some form of NHTSA investigation or recall management at any given time runs into the tens of millions. In a single week ending June 28, the agency processed recalls covering 1.18 million vehicles across five separate actions: Hyundai instrument panels going dark, Toyota EVs losing power at highway speeds, Land Rover airbags that won't deploy, Honda minivan airbags that deploy from potholes, and Tesla steering systems that stop working.[5]
That is one week. On average, NHTSA handles 30 to 40 million recalled vehicles per year. Divide by 555 staff. That is between 54,000 and 72,000 recalled vehicles per employee per year, up from roughly 39,000 to 52,000 before the cuts.
Carl Tobias, a professor at the University of Richmond School of Law who studies regulatory enforcement, summarized the situation with uncomfortable precision: "The short and cute answer is more people will be injured, more people will die."[1]
And Then Congress Proposed Cutting More
A House Appropriations subcommittee advanced a bill that would reduce NHTSA's operations and research funding by more than $10 million and redirect approximately $78 million from the Bipartisan Infrastructure Law, money originally earmarked for state and local safety programs, crash data collection, and behavioral safety research.[6]
Eighty-eight million dollars. For context, NHTSA's entire annual budget is around $1.2 billion, the majority of which passes through to state highway safety grants rather than funding federal investigations. The actual money available for NHTSA to investigate defects, conduct engineering analyses, and enforce recalls against manufacturers with market capitalizations exceeding $700 billion is a rounding error by comparison.
Consumer Reports and a coalition of safety organizations warned Congress that the cuts would weaken "NHTSA's long-term safety capacity" and potentially force "further staffing cuts at an agency that has already experienced losses in key safety and technology areas."[6]
The Fatalities That Frame All of This
Approximately 44,000 Americans die in traffic crashes every year.[7] NHTSA's own FARS database, which we use as the primary source for every story on this site, recorded 278 deaths across 503 fatal crashes involving Tesla vehicles in the 2014 through 2023 reporting period. Those numbers predate the most aggressive FSD deployment, and the Tesla fleet has roughly doubled since. The exposure is accelerating while the oversight apparatus contracts.
We are not claiming that NHTSA staff reductions will directly cause a specific number of deaths. We are saying that a regulatory agency staffed at 555 people, investigating technology deployed across millions of vehicles that operate at highway speeds with varying levels of computer control, has a capacity problem that no amount of "we retained positions critical to the mission" boilerplate can paper over.
Strongest Counterargument
NHTSA's workforce of 555 remains larger than it was for most of its history. In 2023, the agency had approximately 675 employees. Under previous administrations, it operated with fewer. The growth under President Biden, a 30% expansion, was arguably the anomaly, not the current reduction. Further, most recall remedies in the ADAS era are software updates delivered over the air. They do not require the same physical inspection infrastructure as a brake line recall in 1978. NHTSA can, in theory, do more with fewer people because the remedies themselves are more efficient to deploy.
That argument has a structural flaw. Software recalls are easier to deploy, but software defects are exponentially harder to investigate. A brake line fails because of corrosion or a manufacturing defect. A camera-based driving system fails because a neural network misclassifies sun glare as open road. Investigating the latter requires engineers who understand machine learning inference pipelines, sensor fusion architectures, and the statistical methods needed to evaluate probabilistic safety claims. Those people are not interchangeable with the staff who review a brake caliper supplier's metallurgical report, and there were only seven of them to begin with.
Limitations
Staffing figures for the autonomous vehicle team come from Washington Post reporting citing internal sources, not official NHTSA disclosure. The 772-to-555 workforce reduction comes from Reuters citing congressional data; the actual current headcount may differ. The "6.08 million vehicles" figure sums investigation scopes that may overlap, as some Model 3 and Model Y vehicles likely appear in both EA26002 and PE25012. FARS data covers 2014 through 2023, predating the heaviest FSD deployment period. NHTSA may assign non-AV investigators to AV-related casework on an ad hoc basis, meaning the effective AV investigation capacity could be higher than the dedicated team size suggests.
What You Should Do
If you drive a Tesla with any form of Autopilot or FSD: keep your hands on the wheel, treat the system as an assist rather than a replacement for your attention, and check nhtsa.gov/recalls for your VIN quarterly. If you drive any vehicle built after 2015: the same site will tell you whether your car has an open recall, and roughly one in four recalled vehicles never gets fixed because the owner never checked.
If you care about whether the federal government has the capacity to investigate the technology controlling your steering, braking, and acceleration at 70 miles per hour: the House Appropriations Committee's contact page is appropriations.house.gov. Four investigators cannot do this alone. They probably know that better than anyone.
Sources & References
- Carscoops / Reuters, “NHTSA's Shrinking Staff Is Raising Concerns About Safety,” July 2025. Workforce reduction (772 → 555), AV team (7 → 4), Tobias quote. Original Washington Post reporting on the 7-person AV team. carscoops.com
- Reuters / Memeburn, “Tesla Faces Federal Probe After Fatal Texas Crash,” June 2026. EA26002 scope (3.2M vehicles, engineering analysis), PE25012 scope (2.88M vehicles), ~50 special crash investigations since 2016, NHTSA degradation-detection finding. reuters.com
- TechCrunch, “Tesla settles FSD crash lawsuit as federal investigations continue,” June 26, 2026. Johna Story lawsuit settlement, terms undisclosed. techcrunch.com
- Reuters, “US opens probe of fatal Tesla crash into Texas home,” June 23, 2026. Katy, TX crash details, NTSB separate investigation, Elluswamy statement, driver statement. reuters.com
- NHTSA recall filings and Reuters wire reports, week of June 22–28, 2026. Hyundai Tucson (96,310), Toyota/Lexus/Subaru bZ/RZ/Solterra (20,991), Land Rover clockspring (250,857), Honda Odyssey (441,002), Tesla steering (376,241). nhtsa.gov/recalls
- Consumer Reports Advocacy, “Consumer Reports Joins Call to Fully Fund NHTSA in FY26 Appropriations Bill,” July 2025. Budget cut details ($10M operations + $78M BIL redirect). advocacy.consumerreports.org
- NHTSA, Fatality Analysis Reporting System (FARS), 2014–2023. National fatality estimates, Tesla vehicle-specific fatal crash data. nhtsa.gov