Home: Motoring > Tesla FSD v14 Lite Fails to Detect Oncoming Traffic During Nighttime U-Turn in Korea, Safety Driver Intervenes to Prevent Crash

Tesla FSD v14 Lite Fails to Detect Oncoming Traffic During Nighttime U-Turn in Korea, Safety Driver Intervenes to Prevent Crash

From:Internet Info Agency 2026-07-14 11:51:09

On July 13, South Korean automotive testing organization Motline released a 4.5-hour-long video documenting real-world road tests of Tesla’s FSD v14 Lite on Korean roads. One segment showed that during a nighttime U-turn maneuver, the system failed to promptly detect an approaching vehicle rapidly closing in from behind. At the time, the test vehicle was preparing to exit from the U-turn lane, and the rear vehicle was already very close, yet FSD v14 Lite did not respond effectively. The safety driver in the front seat immediately took control of the steering wheel and applied the brakes, averting a potential collision. The test vehicle used was either a Model 3 or Model Y equipped with Tesla’s Hardware 3 (HW3/AI3) computing platform and had an active FSD subscription or purchase. FSD v14 Lite is a lightweight version specifically optimized for the HW3 platform and began rolling out in phases to eligible users in Korea around July 10, 2026, initially targeting Early Access Program participants. This version includes improvements in navigation handling, merging and lane changes, pedestrian interactions, and traffic signal response. The entire test was supervised by a safety driver, with the dashboard continuously displaying map navigation and FSD operational status. Aside from this incident, the system generally performed basic functions such as lane keeping and turning on other road segments. South Korean regulators emphasized that the FSD system remains in a supervised mode, requiring drivers to stay fully attentive and retain full responsibility for driving at all times. Issues identified during such early-stage testing are typically used to inform subsequent software updates and refinements.

Editor:NewsAssistant