NHTSA data from July 1, 2021, to May 15, 2022: six people died in 392 incidents involving driver-assistance tech; Teslas were in 273 crashes and five were fatal
New York Times
Context & Ripple Effects
NHTSA's disclosure gives public incident data a role that experts had said was missing when evaluating Tesla's and other developers' safety claims. It also extends a regulatory record that began with the 2016 Autopilot fatal-crash investigation and the NTSB's finding of driver over-reliance on Autopilot in a separate fatal crash.
The reported concentration of incidents makes Tesla the central company in the first broad NHTSA dataset, rather than leaving driver-assistance safety as a dispute over manufacturer assertions.
First-order effects
Tesla now accounts for most incidents in NHTSA's disclosed dataset, putting its driver-assistance systems at the center of public safety scrutiny.
NHTSA has created a concrete baseline against which Tesla's and other automakers' safety claims can be assessed.
Second-order effects
Researchers, buyers, and regulators gain a common incident record, addressing the previous lack of public data that limited independent checks of autonomous-driving safety claims.
Tesla's crash share raises the stakes for how it demonstrates driver supervision and feature safeguards relative to other automakers reporting incidents.
Third-order effects
If NHTSA continues pairing incident reporting with feature reviews, driver-assistance competition will increasingly turn on evidence of driver engagement; the agency later concluded Tesla's systems did not sufficiently keep drivers engaged in its driver-assistance assessment.
The trend: Driver-assistance oversight is moving from manufacturer safety assertions toward regulator-collected crash evidence and tests of whether systems keep drivers engaged.
In 2019, I investigated the gulf between what Elon Musk would like you to believe about the safety of Tesla's “Autopilot” software and what experts feared to be a dramatic undercount of collisions:https://www.theinformation.c om/ ... Today NHTSA confirmed those fears: https://twi…
This newly disclosed data shows the sweeping scope of dangers in automated driving & driver assistance technology. I've warned repeatedly of these perils & the need to prioritize safety. A lot of work is still necessary but honesty is a first step. https://www.nytimes.com/...
The attention around the @NHTSAgov SGO crash data is yet another clear signal that today's best ADAS is not good enough. There's a huge opportunity to provide consumers with safer, smarter cars & it's our collective responsibility as an industry to act. https://www.nhtsa.gov/...
Lmao all the tesla bros in the comments here like “bUt WhAt AbOuT hUmAn dRiVeRs” yes lads come on you're *so close* to a revelation https://twitter.com/...
This bit is particularly interesting. It's like the autopilot systems wait until the last second, put their fingers to their nose and say, “Not it” https://twitter.com/... https://twitter.com/...
“Teslas operating with Autopilot, the more ambitious Full Self Driving mode or any of their associated component features were in 273 crashes.” https://www.nytimes.com/...
In other news FSD was involved in 250 crashes. In 1.5 billion miles. And it is an exponential learning system so next year the crash number will go down. The year after that it will go down again. Meanwhile human caused deaths have gone up. https://twitter.com/...
this is why supergenius space karen has been trying to distract everybody by behaving like a 4chan toddler for months https://t.co/3hXM4BToO5 https://t.co/O7id7MuxQF
It's actually no surprise that Tesla has the most ADAS crashes (they drive the most miles) but by same token the sheer # of incidents - for a system mainly driven in good weather, well-mapped roads - etc is concerning. & other manufacturers are definitely having issues, too 2/4
Teslas in Autopilot mode have been in 273 crashes in less than a year, according to regulator data, far more than previously known https://www.washingtonpost.com/ ...