Tesla’s Own FSD Trainers Don’t Trust The Technology They Helped Build

Tesla’s Full Self-Driving (FSD) technology faces skepticism from former employees who trained the system, according to a Reuters investigation. Nine former data labelers and a self-driving engineer said they would not trust FSD to drive them, despite years of involvement in its development. Former workers described recurring failures, including Teslas failing to recognize emergency vehicles, school buses, pedestrians, and even children near roadways, as well as striking animals like deer, dogs, and cats without braking. The investigation highlights an internal team nicknamed the ‘trauma team’ due to disturbing footage of near-misses involving pedestrians in crosswalks and residential areas. Former employees also noted inconsistent software behavior, with updates improving some functions while worsening others, and reliability metrics fluctuating unpredictably. Reuters questioned Tesla’s safety claims, which compare FSD-equipped vehicles to broader federal crash data, including less severe incidents requiring only tow trucks. Safety researchers argue this methodology overstates Tesla’s safety advantage, as newer vehicles from other automakers already incorporate advanced crash avoidance technology. The comparison also ignores Tesla’s relatively new fleet against the average American vehicle, which is significantly older. The findings come as Tesla continues to position autonomy and robotaxis as critical to its future, despite growing regulatory scrutiny and lawsuits. Former employees expressed concerns about the technology’s real-world capabilities, undermining public confidence in Tesla’s self-driving ambitions.