Home / Blog / Zoox Recalls Robotaxi Software After a Car Was Confused by Heavy Smoke
Autonomous

🚗 Zoox Recalls Robotaxi Software After a Car Was Confused by Heavy Smoke

By Dillip Chowdary • Jul 17, 2026 • Source: TechCrunch

Amazon-owned Zoox has issued a software recall for its robotaxi fleet after one of its vehicles was thrown off by heavy smoke at an active emergency scene — a real-world edge case that regulators say should never have been an edge case at all.

On June 20, a Zoox robotaxi encountered thick smoke at an uncordoned fire scene, braked hard, and swerved before coming to a stop. A remote teleoperator then reversed the vehicle so first responders could place traffic cones and take control of the area. No one was hurt, and Zoox says it was the only incident of its kind.

The fix is a software update pushed to the company's fleet of 105 vehicles that, in Zoox's words, “enhances the existing capability of detecting active scenes by adding the ability to detect and respond to heavy smoke.” Because the remedy is delivered over the air rather than through a service visit, the “recall” is closer to a patch release than a garage callback — but it is still filed as a formal recall.

The timing is pointed. The recall landed one day before NHTSA administrator Jonathan Morrison publicly warned autonomous-vehicle companies to stop interfering with first responders, framing reliable emergency-scene detection as a functional requirement rather than a rare corner case. For an industry scaling driverless miles fast, “the car got confused by smoke” is exactly the kind of headline it can't afford to repeat.

Key details

  • The incident: June 20: a robotaxi hit heavy smoke at an uncordoned fire scene, braked hard, and swerved.
  • The response: A teleoperator reversed the car so first responders could cordon the scene; no injuries.
  • The fix: An over-the-air update to 105 vehicles adds detection of and response to heavy smoke.
  • Regulatory backdrop: NHTSA's Jonathan Morrison warned AV firms a day later not to interfere with first responders.

Why it matters

Emergency scenes are the ultimate long-tail problem for self-driving systems: rare, chaotic, and high-stakes. How quickly a company detects, discloses, and patches one of these events is becoming the real yardstick for whether robotaxis are ready to scale — and regulators are now watching that yardstick closely.

Source: TechCrunch. Reporting cross-referenced by Tech Bytes on Jul 17, 2026.

🔎 More interesting news

Stay Ahead

Free Tools