
The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.
A ComfortDelGro autonomous shuttle in Singapore collided with a road divider during a routine test after its AI system detected an object and prompted a human operator to take control. Despite the manual intervention, the vehicle crashed, causing property damage but no injuries. Operations were paused pending investigation.[AI generated]
Why's our monitor labelling this an incident or hazard?
The autonomous shuttle is an AI system performing navigation and obstacle detection. The incident occurred during its operation when it detected an object and requested human intervention, but despite the takeover, the vehicle still crashed, causing property damage. This shows the AI system's involvement in the harm, fulfilling the criteria for an AI Incident. There was no injury, but property damage is a recognized harm category. The event is not merely a potential hazard or complementary information, as the harm has materialized.[AI generated]