Tesla’s Full Self-Driving system is facing a deeper level of federal scrutiny, and this time the stakes go well beyond a routine software patch. A fatal crash, a string of related incidents, and mounting questions about how the system reacts to poor real-world visibility have pushed regulators from a preliminary review into a full engineering analysis. That escalation puts one of the industry’s most closely watched technologies squarely in the crosshairs.
For everyday drivers, this isn’t just another autonomous-tech headline. It’s a direct question about whether the system built to assist you can actually perceive what’s in front of it once conditions stop being ideal.
10 Automotive Gadgets Worth Checking Out
What Actually Triggered the Probe
The National Highway Traffic Safety Administration’s Office of Defects Investigation has expanded its review of Tesla’s Full Self-Driving Beta and Full Self-Driving (Supervised) systems, zeroing in on the software responsible for degradation detection, the function meant to recognize when visibility is compromised and alert the driver.
That function became far more important once Tesla dropped radar entirely and committed to a camera-only approach with Tesla Vision starting in mid-2021. The strategy simplified the hardware stack and shifted the burden onto software and vision processing. Regulators are now examining whether that tradeoff holds up in real-world conditions, specifically whether the system properly detects and warns drivers when visibility drops due to glare or airborne obstructions.
Nine Crashes, One Fatality, and a Data Gap
This isn’t a theoretical concern. Regulators are examining at least nine crashes tied to these visibility issues, including one that resulted in a death. Investigators say Tesla began developing a software fix shortly after the fatal crash was reported in late November 2023, but there’s still uncertainty around exactly when that update rolled out and which vehicles received it.
That gap matters more than it might seem. Without clear deployment records, it’s difficult to determine whether the updated software was even active during some of the crashes being reviewed. Regulators believe the software may have played a role in roughly a third of the incidents under investigation.
A Warning System That May Come Too Late
Among the more concerning findings is how the system behaves in the moments leading up to a crash. In multiple reviewed cases, the vehicle either failed to recognize a visibility problem or delayed the driver warning until the last possible moment, a delay that can be the entire difference between avoiding a crash and becoming part of one.
Investigators also documented instances where the system lost track of a lead vehicle entirely, or never detected it in the first place. That’s not a cosmetic bug. Tracking a lead vehicle is a core function of any driver assistance system, and a failure there can escalate a routine drive into a dangerous situation fast.
This Covers a Huge Slice of Tesla’s Lineup
The investigation isn’t narrowed to one model or a tight production window. It spans Tesla vehicles equipped with Full Self-Driving across model years 2016 through 2026, covering the Model S, Model X, Model 3, Model Y, and the Cybertruck. This isn’t a niche concern for a handful of early adopters, it potentially touches a large share of Tesla’s modern lineup, and it raises an immediate question for owners about whether their specific vehicle received the updated software and how well it actually performs.
Why This Matters Beyond Tesla
For enthusiasts, this story is bigger than one brand. It’s about the direction the entire industry is heading. Going camera-only was viewed as a bold simplification, stripping out extra sensors in favor of pure software-driven perception, and that bet may now be under real pressure. If camera-based systems struggle in degraded visibility, it forces a broader industry conversation about whether simplifying hardware comes at the cost of reliability. Drivers who’ve long been skeptical of handing too much control over to software now have a concrete case study to point to.
What Happens From Here
The investigation has moved into an engineering analysis phase, giving regulators room to dig deeper into how the system actually works, how broadly software updates have been deployed, and how effective those updates have been in practice. That process includes reviewing additional incidents and determining whether a broader pattern exists. The outcome could range from a required fix to something far more significant, and for Tesla, the company’s entire approach to driver assistance is now under direct evaluation, with results that could shape how these systems get built going forward.
The Question That Isn’t Going Away
At its core, this comes down to trust. Drivers are being asked to rely on increasingly complex systems to assist, and in some cases take over, critical driving decisions. But if those systems can miss a visibility problem or warn too late, the entire equation changes, because when the technology hesitates or misreads the road, the driver is the one left to react in a split second. The real question is whether the push toward autonomy is outpacing what the technology can currently support safely, and how many more warning signs it will take before that pace changes.