The proposed amendment to FMVSS No. 135, the federal brake standard for light vehicles, would drop the manual brake pedal from purpose built driverless vehicles while leaving stopping distance performance tests intact.
The U.S. auto-safety regulator has proposed dropping the long-standing requirement that self-driving vehicles carry a manual brake pedal, the most concrete federal regulatory shift yet toward a fleet of purpose-built driverless vehicles with no human controls.
The National Highway Traffic Safety Administration filed the proposed rule on June 25 in the Federal Register, targeting Federal Motor Vehicle Safety Standard No. 135, the federal brake standard for light vehicles. The change would remove the mandate that vehicles carry foot- or hand-operated brake controls when they are designed to be operated exclusively by an Automated Driving System, meaning the on-board software drives the car with no human operator. Comments close July 27.
What the rule does not change is more revealing than what it does. Stopping-distance performance requirements remain in force for every vehicle covered by the standard, including driverless ones, validated through alternative test procedures suited to cars with no human driver on board. Vehicles that retain any manual driving controls are still governed by the existing FMVSS No. 135 rule, untouched.
That distinction matters because most public coverage has framed the move as a giveaway to AV operators. The actual mechanism is narrower. NHTSA is replacing a hardware-control baseline with a performance-outcomes baseline: the car still has to stop in the same distances, but it no longer has to prove that by routing its braking through a pedal a human could push. NHTSA Administrator Jonathan Morrison told CNBC the change is meant to tear down "pointless barriers" while preserving "fundamental safety requirements."
The hardware-versus-performance swap is also the answer to a separate problem that has nagged the agency for years. Under current statute, fully driverless vehicles do not need NHTSA approval at all, provided they include the human controls the standard demands: a steering wheel, a brake pedal, mirrors. Manufacturers that want to deploy vehicles without those controls must petition for an exemption, and current law caps those exemptions at 2,500 vehicles per manufacturer per year. NHTSA has been criticized for taking years to act on those petitions.
Zoox's exemption petition is currently in public comment after NHTSA opened a comment period on it in March 2026. General Motors filed a petition in 2018 and withdrew it in 2020, then filed a second in 2022 and withdrew that one in October 2024, citing regulatory uncertainty. The new proposed rule does not retroactively resolve any pending case, but it changes the underlying baseline against which future ADS-only designs will be measured.
The proposal is the fifth FMVSS update under the Department of Transportation's Automated Vehicle Framework. NHTSA is also separately developing real-world safety performance requirements for automated vehicles, a process that will run in parallel with the FMVSS No. 135 amendment.
The two camps around this rule are arguing about what the absence of a replacement framework means. Automakers, frustrated by multi-year exemption reviews and the 2,500-vehicle cap, see a regulator finally catching up to a technology that has been waiting on paperwork. Safety advocates see a softer hardware floor without a replacement visibility tool. The proposed rule itself tries to thread that needle: it does not relax any performance test, and NHTSA retains broad defect-investigation and recall authority over ADS-equipped vehicles regardless of which framework is in force.
What to watch next: the comment window closes July 27, after which NHTSA will need to address material objections before issuing a final rule. Any company currently sitting on an exemption petition will have a clearer picture of its deployment ceiling only after the FMVSS No. 135 amendment and the parallel safety-performance rule both produce final text.