Police says Uber's safety driver in a self-driving vehicle was streaming a Hulu show before the car hit and killed a pedestrian in Arizona in March
SAN FRANCISCO (Reuters) - Police in Tempe, Arizona said evidence showed the “safety” driver behind the wheel of a self-driving Uber …
Context & Ripple Effects
The Tempe investigation has been building since police released dashcam video showing the safety driver looking away repeatedly in the seconds before impact, and since the NTSB found the vehicle itself failed to identify the pedestrian or brake until too late. The new Hulu detail closes the gap between those two findings: it explains what the driver was doing instead of watching the road.
That matters because Uber had already paused all its self-driving tests after the March crash, and the company's entire test protocol rested on a single human backup behind the wheel. Evidence that the designated backup was streaming entertainment converts a system-failure story into an oversight-design story.
First-order effects
- The Tempe safety driver now faces direct criminal exposure, with prosecutors holding evidence of distracted driving at the moment of the fatal collision.
- Uber's one-driver-per-vehicle testing setup is publicly discredited, hardening the case for the suspension already in place across its test cities.
Second-order effects
- Rival AV testers are pushed to add in-cab attention monitoring — cameras and driver-engagement checks — because any single-attendant program now carries the same demonstrated failure mode.
- Insurers and city permitting offices gain a concrete basis to demand recorded driver-attentiveness data as a condition of allowing on-road tests, raising compliance costs across the industry.
Third-order effects
- If prosecution lands on the individual rather than the operator, it sets a template where the 'safety driver' becomes the legally accountable layer — pressuring the industry either to automate the monitoring role out of existence or to formalize it with enforceable standards.
- The pattern points toward regulation treating human oversight claims in autonomous systems as auditable commitments, not assurances, with logged attentiveness data becoming standard evidentiary material in crashes.
The trend: Autonomous-vehicle testing is moving from trust-based human supervision toward instrumented, auditable driver monitoring, as the first fatal crash shifts scrutiny from corporate programs to individual accountability.