July 13, 2026
On July 8, NHTSA Administrator Jonathan Morrison published a letter to the country’s automated-vehicle developers under a title that reads like a verdict: An Automated Vehicle That Cannot Safely Interact With First Responders is a Danger to the General Public. The agency said it has documented a clear pattern of driverless vehicles driving directly into active emergency scenes, blocking the paths of ambulances and firefighters, and failing to recognize flashing lights, flares, smoke, fire, and traffic cones.
Morrison’s letter names no company, and the second reading is the story. It cites no Federal Motor Vehicle Safety Standard, because none covers how a robotaxi should behave at an emergency scene. What NHTSA calls the document is a call to action, and what the agency commits to doing is scheduling meetings: it will sit down with driverless developers by month’s end to hear their solutions. The closest thing to a threat is a sentence promising to continue exercising enforcement authority against developers who do not address significant safety concerns.
The same press release explains why the toolbox is thin, in NHTSA’s own words. Over the past several months the agency has, by its own account, slashed redundant red tape, streamlined the Automated Vehicle Exemption Program and the Part 555 exemption process, and proposed rewrites of the standards governing transmission shifting, defrosting, tire placards, and braking, to the point where individual vehicle exemptions may no longer be necessary. NHTSA also says it has initiated work on the world’s first AV performance standards and the first federal AV safety guidance since 2017. Both are initiated. Neither exists yet.
NHTSA has no first-responder performance standard on the books, and that gap is the whole letter. A regulator with one writes a defect notice. A regulator without one writes a headline and books a meeting. Morrison’s strongest line is the comparison he draws to human drivers, who face fines and jail time for impeding emergency operations, which is precisely the accountability an automated driving system does not have. NHTSA’s open engineering analysis into Tesla’s Full Self-Driving, which GCBC covered in May, is the one live file where the agency does have an instrument, and it took a documented crash pattern to open it.
Morrison closes the letter with a line about public trust being earned rather than given, and none of this makes the document empty. Language like danger to the general public, published by the agency that writes the defect standards, is the vocabulary a regulator uses when it is building a record it intends to use later. The agency is telling developers, in the plainest terms available to it, that the file is open and the next document in it may not be a letter. What it cannot do today is tell them what compliance would look like, because that document has not been written.
Waymo answered the same day, with four more cities
Waymo said on July 8, the same day the letter went out, that it will soon begin fully autonomous operations in San Diego, Las Vegas, Tampa, and Denver, with no human specialist behind the wheel. Those rides start with Alphabet employees before opening to the public, and the four cities join a network Waymo now puts at more than ten. In the same post the company said it has begun driving the Hyundai IONIQ 5 autonomously with a specialist aboard, the validation phase that precedes fully driverless service on a new platform.
Waymo’s expansion is not an answer to the letter, and nothing required it to be. A call to action with no standard behind it changes no deployment plan. The commercial curve and the regulatory curve crossed in the same seven days and only one of them was moving.
Waymo is also the operator with the most developed first-responder program in the industry, including published protocols and a dedicated channel for departments to reach the company, which is worth saying plainly given how the week was reported. The pattern NHTSA describes is not a Waymo pattern by the agency’s own account, and it will not be fixed by any single operator building a better hotline. It is a category problem: an automated driving system has to recognize a scene that is defined by improvisation, where the cones are wherever the fire captain put them.
The states already built what the federal government is asking for
California’s automated-vehicle rules took effect July 1 and let police cite AV manufacturers directly for moving violations, a change GCBC covered when the state also authorized heavy-duty autonomous freight. Sacramento wrote an enforcement mechanism. Washington wrote a letter. When a Waymo or a Tesla robotaxi blocks a fire truck in California, someone can be cited for it, and the federal position is that developers should bring ideas to a meeting.
End of July is eighteen days out. Tesla reports second-quarter results on July 22 and will be asked about the Cybercab, a vehicle with no steering wheel that needs an exemption from the same agency now describing the category as a public danger. Whatever the developers put in front of NHTSA before August, the useful question for readers is not whether the meetings happen. It is whether the competency standards NHTSA says it has initiated arrive before the next ambulance gets blocked.
