US Senators Demand NHTSA Review of Autonomous Vehicle Safety Claims Amid Data Transparency Push

The promise of autonomous driving has always rested on a single, powerful premise: computers do not get distracted, they do not get tired, and they can ultimately drive safer than humans. As automakers and tech companies scale their driver-assistance and robotaxi fleets across the United States in 2026, the industry is leaning heavily on statistics to prove this point to a skeptical public. Tesla, in particular, has aggressively promoted its "Full Self-Driving" (FSD) software, with executives frequently claiming the system makes its vehicles up to ten times safer than human-driven cars [1, 6]. But as these systems transition from beta tests to widespread deployment, the math behind those safety claims is facing unprecedented scrutiny from federal lawmakers and independent researchers.[1]

On June 16, 2026, United States Senators Edward Markey and Richard Blumenthal formally demanded that the National Highway Traffic Safety Administration (NHTSA) investigate the data underlying Tesla's safety assertions [6]. In a pointed letter to NHTSA Administrator Jonathan Morrison, the lawmakers characterized the automaker's self-published statistics as "weak and misleading," warning that the company's data practices have created an urgent safety problem on public roads [1, 2]. The senators argue that inflated safety statistics can encourage drivers to over-rely on systems that still require active human supervision, obscuring potential defects and undermining regulatory oversight [6].[1][2]

The political pushback stems from a broader investigation into how automakers calculate the crash rates of their advanced driver-assistance systems (ADAS). According to traffic safety researchers, the headline figures often rely on methodological choices that systematically inflate the apparent safety advantage of the software [3, 5]. The core issue is an "apples-to-oranges" comparison that measures unlike crash outcomes against each other, creating a distorted picture of how autonomous systems perform in everyday driving conditions [2, 6].[2]

The first major methodological discrepancy involves crash severity thresholds. When calculating the crash rate for vehicles operating on FSD, Tesla allegedly only counts severe accidents that are violent enough to trigger an airbag deployment [1, 7]. However, this figure is then compared against the overall national crash rate for all vehicles, a broad federal dataset that includes minor fender-benders, scrapes, and low-speed collisions [1, 7]. By filtering out minor incidents from its own data while keeping them in the national baseline, the resulting comparison mathematically guarantees a dramatically lower crash rate for the software.[1][3]

The second structural flaw identified by lawmakers involves the age and technological baseline of the vehicles being compared. New electric vehicles equipped with advanced driver-assistance systems are inherently packed with modern passive and active safety features, from advanced crumple zones to automatic emergency braking [1, 7]. When these state-of-the-art vehicles are compared to the average American car—which is currently around twelve years old—the newer vehicles naturally exhibit better safety outcomes [4, 6]. Critics argue this baseline advantage is being improperly credited to the software itself, rather than the underlying modern hardware.[1][3]

Perhaps the most consequential data point under scrutiny is the "disengagement window"—the precise moment a human driver takes control back from the software right before a collision. Under NHTSA's Standing General Order, a crash is attributed to an automated system if the software was engaged at any point within thirty seconds prior to the impact [3]. However, lawmakers have questioned whether Tesla only counts a crash against FSD if the system was active within five seconds of the collision [3, 4]. If a driver realizes a crash is imminent and grabs the wheel six seconds before impact, that incident might vanish from the software's safety record entirely.

The implications of these data practices extend far beyond domestic regulatory debates. Tesla recently utilized these same self-published safety figures in its submissions to European regulators, seeking approval to deploy FSD across the continent [3, 7]. While the Dutch road authority, RDW, approved a supervised version of the software in April 2026, the agency noted that it relies on its own independent testing and verifications rather than marketing claims [3, 7]. Nevertheless, the cross-border use of contested statistics highlights the global need for standardized autonomous vehicle metrics.[3]

The push for transparency also touches on how data is shielded from public view. Lawmakers have criticized the practice of redacting large portions of safety data submitted to regulators under the umbrella of "confidential business information" [4]. When critical crash metrics are hidden from independent researchers and the public, it becomes nearly impossible for third-party safety organizations to peer-review the claims being made in consumer marketing campaigns.

Beyond scrutinizing a single automaker, the senators' letter reveals a significant gap in the current federal regulatory framework. Markey and Blumenthal are urging NHTSA to drastically tighten reporting rules for all companies deploying advanced driver-assistance systems and autonomous technology [1, 5]. Under the current guidelines, automakers are required to report relatively little raw safety data, leaving the agency with no concrete way to verify whether public safety claims bear any relationship to reality [1, 6]. The lawmakers are pushing for a standardized data-collection mandate that would force all companies to use identical metrics.[1]

This is not the first time these specific lawmakers have targeted the intersection of software and road safety. In late 2025, Senators Markey and Blumenthal introduced the "Stay in Your Lane Act," legislation designed to prohibit self-driving technology from operating outside the specific road conditions and environments for which it was originally designed and tested [6]. Their ongoing legislative efforts reflect a growing frustration on Capitol Hill with the pace at which federal regulators are adapting to software-defined vehicles.

The financial markets are also closely watching how regulatory scrutiny might impact the rollout of these highly profitable software packages. Following the release of the senators' letter, shares of major electric vehicle manufacturers experienced slight volatility, though broader delivery forecasts for the second quarter of 2026 remained strong [4]. Wall Street understands that the widespread monetization of autonomous software depends heavily on maintaining regulatory goodwill and avoiding mandatory recalls or feature restrictions.

The distinction between different levels of autonomy makes this data transparency even more critical. Systems like Tesla's FSD and General Motors' Super Cruise are classified as SAE Level 2, meaning they require a human driver to remain attentive and ready to take control at all times [2, 6]. In contrast, Level 4 systems, such as the robotaxis operated by Waymo, are designed to drive themselves without human intervention under specific conditions [2, 5]. When Level 2 systems are marketed with statistics implying superhuman safety, regulators worry that drivers will treat them like Level 4 systems, leading to dangerous complacency.[2]

NHTSA, which already has multiple active investigations into various driver-assistance systems, has confirmed receipt of the senators' letter and is currently reviewing the demands [1, 7]. In March 2026, the agency upgraded one of its probes into advanced driver-assistance systems to an Engineering Analysis—a more advanced investigative stage that covers millions of vehicles and often precedes a formal recall [4]. This existing regulatory pressure adds significant weight to the senators' request for a comprehensive data audit.[1][3]

The agency has been given a deadline of July 7, 2026, to answer a series of pointed questions, including whether it has ever independently evaluated the specific safety claims in question or demanded the raw, unredacted crash data behind them [1, 3]. The response will likely signal how aggressively the federal government plans to police the marketing of autonomous technology, and whether it will mandate a unified reporting standard for the entire industry. For the automotive sector, the transition from corporate marketing claims to standardized, independently verified empirical evidence will be the defining hurdle for the future of transportation.[1]