Priya works nights in a Pittsburgh emergency room. At 3 a.m., when her shift ends and the buses have stopped running, she orders a robotaxi home. She trusts it enough to close her eyes for a few minutes. But she has never seen proof that it met real self-driving safety standards before regulators let it carry her.
That gap just got smaller. On September 22, the National Highway Traffic Safety Administration announced a new partnership with SAE International. SAE already writes the rulebook automakers follow when they engineer new cars. Together the two launched A2SCEND, a consortium built to write the first real, measurable competency test for self-driving vehicles nationwide. Right now, no such test exists. Companies largely grade their own homework.
What the New Self-Driving Safety Standards Actually Cover
A2SCEND brings AV makers, fleet operators, first responders, and safety regulators into one room instead of fifty separate ones. NHTSA Administrator Jonathan Morrison announced the effort at the AI Horizons Pittsburgh Summit, and he was blunt about why it matters. People will only ride in these cars if they trust them, he said, and that trust “is not guaranteed.”
Regulators wrote today’s federal safety rules for cars with steering wheels and pedals. Tesla’s Cybercab and Zoox’s robotaxi have neither. Regulators have been fitting square vehicles into round rules. A2SCEND’s first job is different. It will build a standardized “ADS competency” exam, similar in spirit to a driving test. Any autonomous system will have to pass it before earning a federal safety rating.
The Emergency Vehicle Problem This Is Meant to Fix
This isn’t an abstract policy exercise. This July, dozens of Waymo vehicles got stuck in San Francisco gridlock during a holiday celebration. They failed to clear the way for emergency crews. Robotaxis have also driven past stopped school buses in several cities. NHTSA’s updated guidance feeds directly into A2SCEND’s work. It now requires AV companies to respond to a first responder’s hand signals or flares within 30 seconds. It also lets local emergency officials geofence robotaxis out of active incident scenes.
That’s a real safety win for anyone who has ever needed an ambulance to get through. It also matters if you simply live near a fire station.
What This Could Actually Save You
Here’s the part I care about most: a single national test replaces a patchwork of state-by-state rules that slows everything down. Every extra month a company spends satisfying fifty different paperwork processes costs you something. It’s a month your city waits for affordable late-night rides. It’s a month your local transit agency delays the wheelchair-accessible pilot it keeps promising. NHTSA and SAE are funding A2SCEND with $5 million over three years, a modest sum next to what fragmented compliance costs the industry today. TechCrunch’s mobility desk put it well: without a shared yardstick, nobody, including regulators, can say with confidence when an AV is actually safe enough.
“This future is not guaranteed. They’ll only adopt it if they trust it.”
Jonathan Morrison said that, and he runs the agency writing the rules. It’s worth taking seriously.
Germany and the EU are racing to write their own versions of this test, and China already published its own crash-cutting safety standard. Whichever standard earns public trust first could end up shaping the cars on your street for a decade. Watch for A2SCEND’s first published benchmarks. That’s the moment self-driving safety stops being a marketing claim and starts being a number you can actually check.