Former Tesla Employees Question Autopilot Reliability and Safety Data Integrity

Several former data annotators and engineers who previously worked on Tesla's AI training have expressed distrust in the company's Autopilot system and its publicly released safety data. Interviewees pointed out that the system struggles in specific scenarios—such as yielding to emergency vehicles—and noted that Tesla makes the system appear more reliable by repeatedly training it on the same situations over extended periods. Other experts argue that Tesla’s crash data lacks validity and could constitute misleading marketing. More than seven former employees voiced skepticism about the system, citing unusual incidents including high-speed collisions with animals and instances of excessive speeding.