The National Highway Traffic Safety Administration (NHTSA) found that 11,704 vehicles – distributed across Tesla’s Model S, X, 3 and Y range – were estimated to have the defect.
The glitch was actually introduced by Tesla itself after it released a firmware update to the vehicles (2021.36.5.2) on October 23 2021 that gave drivers limited early access to its version 10.3 ‘Full-Self Driving’ (FSD) beta software.
However, the update introduced a software communication disconnect between the two onboard chips; specifically, when the vehicle is waking up from ‘Sentry Mode’ or ‘Summon Standby Mode’.
This communication disconnect could result in the video neural networks that operate on that chip running less consistently than expected, the NHTSA said. The rollout of the new software was quickly halted while a fix was found and new software was deployed in an over-the-air update.
NHTSA said Tesla “uninstalled FSD 10.3 after receiving reports of inadvertent activation of the automatic emergency braking system” and then “updated the software and released FSD version 10.3.1 to those vehicles affected.”
After reports about the glitch first started appearing on October 24, Tesla founder Elon Musk tweeted that same day: “Seeing some issues with 10.3, so rolling back to 10.2 temporarily. Please note, this is to be expected with beta software.”
Tesla’s FSD software became a controversial feature even before it reached consumers. In October, the US National Transportation Safety Board (NTSB) – another US regulator – urged Musk’s firm to limit where the feature could be used over concerns that it was not safe enough to use on roads in every circumstance.
The NTSB is still investigating a fatal accident from April this year that happened while a Tesla was in self-driving mode and which killed both occupants of the vehicle.
In July, safety experts said that the self-driving software “lacks safeguards” after they tested its capabilities and found it was scraping against bushes, missing turnings and even heading towards parked cars.