NHTSA Moves to Set Objective Safety Standards for Autonomous Vehicle Behavior
The National Highway Traffic Safety Administration is drafting objective safety standards for how self-driving cars behave on the road, moving from one-off incident enforcement to competency-based rules. Jonathan Morrison, NHTSA’s top official, told Bloomberg in an interview published July 16, 2026, that the goal is to craft requirements “so that a manufacturer who is producing the vehicle can know with certainty whether or not the vehicle meets the requirements.” The agency plans to seek public and industry comment first, with the aim of identifying “behavioral competencies” and tests that measure them, and officials want the standards finalized before the end of President Donald Trump’s current term.
AV exposure is currently measured the way you measure something you cannot inspect: by counting what has already gone wrong. Incident reports and recall patterns are lagging indicators, and they say more about a manufacturer’s fleet size and disclosure practices than about how its vehicles behave. The behavioral data that would answer the question directly sits with the manufacturer, which is the same access problem carriers hit with telematics in conventional auto. A competency standard would give underwriters something they have not had, which is a federal benchmark that a manufacturer either meets or does not, independent of how many vehicles it has on the road to generate incidents in the first place.
What counts as a competency is still open, and the comment request NHTSA says it will put out first is where that gets settled. It has not gone out yet. One candidate is already on the table anyway: days before the interview, Morrison publicly pressed AV developers to fix their vehicles’ interference with first responders, an ordinary-road scenario that human drivers resolve without instruction. Carriers writing commercial auto or fleet coverage for AV operators have a reason to be ready to file the day it opens rather than wait for the final rule, because the competencies selected now are the ones underwriters will be asking manufacturers to document later.
NHTSA said it would keep addressing individual safety incidents through its recall and enforcement authorities, with the competency standards setting broader parameters around them. That dual track is the operative detail for underwriters, because a manufacturer can hold a clean certification and still accumulate recalls, and nothing in the framework forces those two records to agree. Pricing AV risk will mean reading both records, the same two-source problem showing up across every line as AI enters the underwriting file. Underwriters who expect to lean on the certification should assume the recall docket will contradict it at some point, and settle in advance which one moves the rate.