Forty-three videos of people apparently asleep at the wheel of a moving Tesla have become a matter for the U.S. Congress. Raja Krishnamoorthi, a Democratic Representative from Illinois, wrote to Transportation Secretary Sean Duffy on September 9, demanding an account of the actions taken by the Department of Transportation and the National Highway Traffic Safety Administration regarding Tesla's Autopilot and Full Self-Driving systems.
The letter draws on an NBC News investigation, which reportedly identified 43 videos posted online: 17 of which allegedly date back to 2026 alone. In themselves, they are neither technical proof of how each individual car operated nor do they allow for reconstructing all driving conditions. However, for the lawmaker, they outline behavior widespread enough to demand a response from federal authorities, especially since Tesla explicitly states that its systems require active driver supervision and do not constitute autonomous driving.
The case comes as NHTSA has already opened an engineering analysis into approximately 3.2 million Tesla vehicles equipped with FSD. The investigation, identified as EA26002, is in the engineering analysis phase: the most advanced stage of the agency's investigative process, which can lead either to closure or to a formal recall request. According to the elements cited in the letter, the inquiry covers nine crashes that occurred in conditions of reduced visibility, resulting in one fatality and one injury.
The issue is not just about those who break the rules
The point raised by Krishnamoorthi is not that Tesla can make the conduct of a driver sleeping at the wheel lawful: it cannot, and the company's instructions require continuous attention. The issue is whether a system sold as driver assistance, but used in practice as a handover of driving itself, has sufficiently effective safeguards against foreseeable misuse.
For the automotive industry, this distinction is crucial. Advanced driver-assistance systems, often referred to by the acronym ADAS, operate within a framework in which the driver remains responsible for the vehicle. However, their operation can induce overconfidence, especially when they manage steering, acceleration, and braking for extended periods. If the car appears capable of handling a stretch of highway or urban driving autonomously, some users may end up lowering their level of vigilance to the point of entirely disengaging from the driving task.
Monitoring driver attention therefore becomes a safety feature just as important as external sensors and the software controlling the vehicle's progress. Tesla uses an in-cabin camera to check the behavior of the person in the driver's seat. However, the letter contends that simple methods exist to circumvent its effectiveness, citing examples such as obstructions placed in front of the camera or sunglasses used to make eye tracking less reliable.
This is not a minor detail. A control system designed to ensure the driver is watching the road must be evaluated not only under ideal conditions, but also regarding the possibility of intentional evasion. Moreover, videos circulating online demonstrate an aspect automakers and regulators know all too well: dangerous behaviors can be filmed and posted as proof of the vehicle's capabilities, amplifying practices strictly prohibited by user manuals.
The drowsiness alert in the crosshairs
In his letter, Krishnamoorthi also points to a warning displayed in Tesla vehicles: “Drowsiness detected. Stay focused with FSD”—that is, the detection of drowsiness coupled with an invitation to stay focused using FSD. For the congressman, the phrasing is problematic because it links the system to the exact condition in which the driver may be least capable of supervising it.
The criticism concerns the interaction design, not just the words on the screen. When the car detects signs of fatigue, a simple prompt to pay attention may not be an adequate measure if the driver is already on the verge of falling asleep or has lost the ability to intervene quickly. The alternative is not defined in the letter and it is up to the authorities to evaluate it within the applicable regulations, but the question touches on the boundary between an informational warning and effective protection.
It is also one of the most delicate issues for Tesla. The name Full Self-Driving can be perceived by the public as a promise of complete autonomy, while the company specifies that the system does not transform the car into an autonomous vehicle and requires constant supervision. The gap between commercial branding, social media messaging, operating instructions, and real-world user behavior has been at the center of the debate over driver-assist system safety for years.
The responses awaited from the Department of Transportation
The congressman asked Duffy to respond by September 30. The requests concern the agency's awareness of the videos, the possible existence of a systemic problem, contacts with Tesla regarding the reported cases, and data collected on driver distraction crashes since January 1, 2024. Krishnamoorthi also wants to know how NHTSA assesses the effectiveness of Tesla's internal camera and what evidence could prompt the agency to mandate a safety recall.
- whether the footage has already been examined by federal authorities;
- whether driver monitoring is considered adequate against inattention and circumvention;
- what data and thresholds could lead to formal action on the vehicles involved.
A letter from a member of Congress does not automatically alter an investigation or force Tesla to change its software. However, its significance lies in increasing public pressure on the DOT during an ongoing proceeding. For NHTSA, the issue is not limited to the conduct of a few users filmed online: it will have to determine whether available data indicates a defect or an unreasonable risk related to the design, intended use, and reasonably foreseeable misuse of the system.
A potential recall would not necessarily involve a mechanical intervention. In the connected vehicle sector, many safety campaigns rely on software updates, modifications to in-cabin alerts, limitations on available features, or stricter monitoring requirements. However, engineering analysis EA26002 has not yet yielded a public conclusion of this kind, and it is impossible to anticipate the outcome of the investigation.
For Tesla, the confrontation has consequences that extend beyond this single dossier. The ability to demonstrate that a driver remains genuinely engaged in driving is an operational and reputational requirement for any advanced assistance feature. For the American regulator, on the other hand, the case serves as a testing ground for how effectively existing rules and verifications can keep pace with cars that assume an increasing share of the driving task, without relieving the person behind the wheel of ultimate responsibility.
By September 30, attention will therefore turn to the DOT's responses. The decisive phase will then remain open: determining whether the agency will view the footage and crash data as elements of a structural problem, or rather as manifestations of misuse contrary to instructions that current systems are not required to prevent beyond a certain threshold. It is along this line that the viability of the supervision model adopted by Tesla will be measured.
