NHTSA Cleared 1 Million Teslas in One Week. The Same Week, FSD Killed Someone.

Between June 27 and July 2, NHTSA closed two separate Tesla safety probes covering 1,071,241 vehicles, with no enforcement action in either case. Six days, seven digits of cleared cars, zero consequences for the manufacturer whose own engineering decisions caused the defects under investigation.

On June 27, the agency shuttered Engineering Analysis EA24001 into power steering loss in 376,241 Model 3 and Model Y vehicles.^[1] The root cause was an overvoltage breakdown that fried motor drive components on the electronic power steering circuit board. Tesla pushed an over-the-air patch in October 2023 and filed a formal recall fifteen months later, in February 2025, which it stressed was "not in response to NHTSA's investigation." Translation: we fixed it because we felt like it, not because you told us to. NHTSA's response to that posture was to close the case.

Five days later, on July 2, NHTSA closed Preliminary Evaluation PE22002 into unexpected deceleration in 695,000 Model 3 and Model Y vehicles from the 2021 and 2022 model years.^[2] This one had been open since February 2022, which means the agency spent four years and four months investigating phantom braking events that, by its own admission, never caused a single crash. At its peak the probe tracked 300 complaints; by 2026, that number had fallen to three.

The strongest case for NHTSA's decision is straightforward: Tesla shipped software updates, and they worked. Three complaints across a 695,000-vehicle fleet is a vanishing failure rate. Regulator found a problem, manufacturer fixed it, complaints collapsed. That is the system functioning exactly as designed. If every open probe ended this way, NHTSA's track record would be exemplary.

NHTSA's closure document identified the probable cause: Tesla's mid-2021 decision to rip out forward-facing radar and rely exclusively on camera-based vision sensing.^[3] That architectural gamble, driven by Elon Musk's conviction that cameras alone could replicate human perception, generated thousands of phantom braking events across highway driving, where drivers reported sudden 10–20 mph decelerations lasting one to three seconds with no obstacle present. The agency found no lateral lane departure and no collision risk significant enough to act on. It acknowledged that "closing this investigation does not constitute a finding that the safety-related defect does not exist." That sentence is doing extraordinary rhetorical work: we found the defect, we know what caused it, we're leaving.

One day after that closure, on July 3, prosecutors in Katy, Texas, charged a Tesla driver with manslaughter for crashing a Full Self-Driving vehicle into a home and killing Maria Avila.^[4] The driver, Butler, had been searching Google for "Tesla fsd not aggressive enough 2026" and "Tesla fsd too timid" before the crash. FSD was steering the entire time; Butler overrode the speed with the accelerator, which is exactly what the system invites drivers to do when its "Sloth" profile creeps below the speed limit. He told paramedics he "passed out."

This is the regulatory paradox at the center of Tesla's safety record, reduced to a single week in late June and early July 2026: NHTSA clears the cars on Tuesday, the courts charge the drivers on Thursday. Its methodology runs on complaint volume: complaints go down, case goes away. But complaints decline for structural reasons that have nothing to do with safety: owners sell the car, software masks the symptom, fleet ages out, or people simply stop reporting to an agency that spent four years investigating and did nothing. NHTSA has opened roughly 50 special crash investigations into Tesla vehicles since 2016, documenting approximately two dozen deaths.^[5] It keeps investigating, and it keeps closing. The $243 million jury verdict in Florida, which found Tesla 33% liable even when the driver was clearly misusing Autopilot, suggests the courts have developed a more durable framework for accountability than the regulator tasked with providing it.

Limitations matter here. This article connects three legally distinct events that happened to share a calendar week: two routine probe closures and a state criminal prosecution. NHTSA cannot weigh pending criminal cases in its engineering analyses, and the manslaughter charge against Butler is procedurally independent from the agency's decision to close either probe. Complaint-decline is one possible explanation for the closures; it is not the only one, and NHTSA's full analytical record is not public.

What you should know: If you drive a 2021–2023 Model 3 or Model Y, your vehicle was covered by at least one of these probes. The phantom braking investigation found no crash risk, and the power steering fix has shipped via OTA. But NHTSA's own documents identify Tesla's engineering decisions as the cause of both defects. "Closed" does not mean "safe." It means the agency looked, found what it found, and chose not to act. Check your vehicle's recall status at nhtsa.gov/recalls and verify that all applicable software updates have been applied.