A driver slumped behind the wheel while a car continues down the highway is the kind of image that instantly cuts through technical arguments about automated driving. A growing collection of videos appearing to show Tesla drivers asleep or seriously disengaged has now turned that image into a fresh regulatory problem.
The controversy is not simply about whether Tesla’s technology can keep a vehicle moving while its driver looks away. Autopilot and Full Self-Driving (Supervised) are explicitly designed around continued human supervision. The harder question is whether their safeguards are strong enough when drivers misuse systems that have become capable of handling long stretches of driving. With U.S. lawmakers demanding answers and Canadian police issuing tickets, that question is moving well beyond social media.
Dozens of Videos Have Turned Misuse Into a Regulatory Issue
The latest pressure intensified on September 10, when U.S. Representative Raja Krishnamoorthi urged the Department of Transportation and National Highway Traffic Safety Administration to act over apparent misuse of Tesla’s Autopilot and Full Self-Driving systems. He cited an NBC News investigation that identified 43 videos appearing to show people asleep in the driver’s seat of moving Teslas. Seventeen of those videos were posted during 2026. That is a large enough collection to raise a different question from an isolated viral clip: whether predictable misuse is recurring despite the safeguards already built into the vehicles.
Krishnamoorthi asked regulators whether they have assessed the effectiveness of Tesla’s driver-monitoring system when cameras are obstructed or when drivers wear sunglasses, hats or other items that could interfere with monitoring. He also asked what conditions would cause NHTSA to regard the monitoring system as a safety defect serious enough to require corrective action or another recall. Reuters reported on September 10 that Tesla, NHTSA and the Transportation Department did not immediately respond to its requests for comment. Krishnamoorthi requested written answers by September 30.
Tesla Says Neither Autopilot Nor FSD Replaces the Driver
Part of the confusion surrounding the videos comes from what Tesla vehicles can visibly accomplish. Autopilot can manage functions such as steering, acceleration and braking under appropriate conditions, while FSD (Supervised) can perform a broader set of tasks including lane changes, turns and route navigation. From another vehicle, a Tesla travelling smoothly down a highway can therefore look remarkably self-sufficient. Yet Tesla’s current documentation draws a firm line between advanced assistance and autonomy.
Tesla says FSD (Supervised) does not make a vehicle autonomous and requires a fully attentive driver who can take over at any time. The company warns owners not to become complacent. At the same time, Tesla argues that the technology delivers substantial safety benefits. Its current Vehicle Safety Report claims vehicles operating with FSD (Supervised) experience seven times fewer major and minor collisions and five times fewer off-highway collisions than the estimated U.S. average. Those are Tesla’s own fleet statistics, and the company still describes FSD as supervised technology. In other words, even Tesla’s safety case assumes an attentive human remains part of the system.
A Car Continuing to Drive Is Not Proof That Sleeping Is Safe
The disturbing feature of many sleeping-driver clips is precisely that nothing dramatic seems to happen. The car stays in its lane, follows traffic and continues moving. That apparent normality can create a misleading impression that the system successfully handled the situation. In reality, a supervised driving system remaining stable for several minutes does not demonstrate that it could safely deal with every sudden event during that period. An unexpected obstruction, unusual road geometry, emergency vehicle, construction zone or sensor problem could still require immediate human intervention.
There is another important limitation when interpreting online footage. A person filming from another vehicle generally cannot establish which exact Tesla assistance mode is active unless the interior screen or other evidence is visible. Some videos cited by lawmakers and journalists reportedly show FSD operating, while other clips mainly establish that the Tesla was moving while the driver appeared disengaged. That distinction matters for technical accuracy. What the broader collection does demonstrate is a behavioural problem: some drivers appear sufficiently confident in Tesla assistance technology to stop performing the supervisory role that the manufacturer says remains mandatory.
Tesla Has Added Stronger Driver-Monitoring Safeguards
Tesla’s newer systems do considerably more than the early steering-wheel reminders associated with Autopilot. Current FSD documentation says the cabin camera monitors attentiveness and that the monitoring function cannot be disabled. Tesla specifically states that the system remains active when a driver wears sunglasses. If it detects repeated attention lapses, the vehicle displays escalating warnings. Depending on what the cabin camera can see, it can also request slight steering-wheel force as another indicator that the driver is available to take control.
Persistent failure to respond can have progressively stronger consequences. Tesla says FSD can be disabled for the remainder of the drive after repeated ignored warnings. If the driver still does not take over, the vehicle can sound a continuous warning, activate its hazard lights and pull over. Tesla also warns that devices intended to defeat attentiveness safeguards violate its terms and can result in loss of access to the feature. Those measures show that the company clearly recognizes disengagement as a foreseeable risk. The current controversy is about whether those protections consistently identify real-world behaviour quickly enough, particularly when users intentionally experiment with their limits.
NHTSA Has Already Recalled Teslas Over Foreseeable Autopilot Misuse
The sleeping-driver debate is landing on top of years of federal scrutiny rather than starting a new regulatory chapter from scratch. In December 2023, Tesla recalled 2,031,220 U.S. vehicles equipped with Autosteer after NHTSA determined that, in some circumstances, the scope and prominence of the system’s controls might be insufficient to prevent driver misuse. Tesla responded with an over-the-air software remedy intended to strengthen warnings and engagement controls.
The investigation behind that recall examined hundreds of crashes. NHTSA’s closing report recorded 467 crashes in its dataset, 54 injuries and 14 fatalities. The regulator said foreseeable driver misuse had played an apparent role in at least 13 crashes involving fatalities. NHTSA subsequently opened a Recall Query to examine whether Tesla’s 2023 remedy was adequate after identifying post-remedy crashes and concerns during testing. That history is particularly relevant to the newest videos. Federal regulators have already formally accepted the principle that safety design must consider foreseeable misuse. The argument therefore cannot end simply by noting that drivers are breaking Tesla’s rules; regulators can also ask whether the product does enough to stop them.
Human-Factors Research Explains Why Capable Automation Can Encourage Risk
There is a difficult psychological problem built into supervised automation: the better a system performs ordinary driving, the less work the human appears to have to do. Yet that same human may suddenly be needed when the system encounters something unusual. Research has repeatedly found that distraction and reduced monitoring can damage takeover performance. In one on-road study involving 36 participants driving a Tesla Model S with Level 2 automation, researchers found extremely long off-road glances during high-distraction conditions, occasionally exceeding 30 seconds. Being eyes-off-road before an unexpected takeover request produced more delay than simply having the hands off the steering wheel.
A later naturalistic study followed 14 drivers using Tesla Autopilot for a month. Researchers found that drivers gradually learned how the safeguard sequences behaved. They responded more quickly to warnings, but also spent more time performing non-driving activities and looking away from the road as they gained experience. The study recorded 16 automation lockouts. That does not mean every driver will abuse assistance technology. It does show why safety engineers worry about systems whose human supervisor can learn exactly how much disengagement the safeguards tolerate.
British Columbia Is Already Seeing the Problem on Real Roads
For Canadian drivers, the issue is not theoretical. On August 11, B.C. Highway Patrol received reports from two witnesses who said a woman appeared to be sleeping while driving a Tesla on Highway 1 near North Vancouver. A passenger recorded video, police traced the vehicle to its registered owner and a 45-year-old woman was issued a $368 ticket for driving without due care and attention. Police stressed that self-driving technology did not excuse the driver from remaining alert.
Another incident occurred on August 27 near Langley. An officer travelling to a priority call spotted a Tesla during morning rush hour whose driver appeared to have his eyes closed and both hands off the wheel. The officer diverted from the original assignment to stop the car. Police said the body-worn camera appeared to show the driver asleep, although the driver denied sleeping. He received a $368 due-care ticket and an additional $81 fine for failing to produce a driver’s licence. B.C. law also separately prohibits Level 3, Level 4 and Level 5 automated vehicles on public highways unless regulations provide otherwise. Current supervised Tesla assistance does not convert the driver’s legal role into that of a passenger.
FSD Is Already the Subject of Two Major U.S. Safety Investigations
Sleeping drivers add another concern to a regulatory docket that is already crowded. In March 2026, NHTSA escalated an investigation covering approximately 3.2 million Teslas equipped with FSD after examining the system’s performance when visibility deteriorates. The agency said it had reports of nine incidents potentially connected to conditions such as glare, dust and other airborne obstructions, including one fatal crash and two injury crashes. Investigators expressed concern that Tesla’s camera-based system might not always identify deteriorating visibility or provide sufficient warning before a crash.
A separate investigation, opened in October 2025, covers about 2.88 million vehicles equipped with FSD after more than 50 reports involving potential traffic-safety violations. NHTSA said it had received reports involving Teslas running red lights and moving against the proper direction of travel during lane changes. The relevance to sleeping-driver cases is straightforward. A supervised system is designed on the assumption that a person will catch situations the software mishandles. If that person is asleep, watching a screen or otherwise disengaged, the final layer of redundancy disappears at exactly the moment it may be needed.
Tesla’s Drowsiness Messaging Is Now Part of the Debate
Lawmakers are also scrutinizing how Tesla talks to a driver before FSD is engaged. A 2025 software update was reported to include messages such as “Drowsiness detected. Stay focused with FSD” and a separate prompt suggesting FSD after lane drift. The messages attracted attention because Tesla simultaneously requires continuous supervision while FSD is operating. WIRED reported that human-factors researchers questioned whether reducing a tired driver’s physical driving workload could make it easier for that person to become even less engaged.
Krishnamoorthi specifically raised that issue in his September letter. His concern was not that driver assistance can never help an alert person after a momentary mistake. The harder scenario is a system detecting possible drowsiness and then encouraging activation of another system that depends on that same driver remaining continuously ready to intervene. Tesla’s underlying argument may be that FSD can help stabilize the driving task when human performance declines. The regulatory question is whether that benefit is outweighed by the possibility that the prompt reinforces an already tired driver’s belief that the car can safely take over.
The Next Fight Is About Where Responsibility for Misuse Ends
No serious interpretation of these incidents removes responsibility from drivers who deliberately sleep or perform other distracting activities. Tesla’s instructions are explicit, and police can ticket drivers who surrender attention. What regulators must decide, however, is how much responsibility manufacturers have when misuse is predictable, repeated and potentially encouraged by the capability of the product itself. NHTSA already used the phrase “foreseeable driver misuse” in its earlier Autopilot investigation, making that distinction especially important.
Krishnamoorthi’s latest letter asks whether the pattern of public videos indicates a systemic weakness in Tesla’s driver-engagement protections and whether regulators have enough crash and driver-state data to judge the problem properly. NHTSA can request information, conduct defect investigations and, when evidence supports a safety defect, pursue corrective action that may include a recall. That does not mean another recall is inevitable. It means the sleeping-driver videos have arrived at a sensitive point: Tesla’s assistance technology is becoming more capable, the company is pushing deeper into autonomous vehicles, and regulators are being asked whether supervision safeguards have kept pace with the confidence some drivers place in the software.
