All Google self driving cars have a human driver who can take control at any time. I don't know if the human was driving the car when this accident occurred, but so far there is no evidence supporting the notion that Google is lying to the cops (which would be a crime).

> Why would a human driver be driving an automated car?

Because Google cannot legally operate the car in the streets without a driver. Therefore they put a driver there and it's on him to choose how to drive the car.

> What is the security mechanisms that allow such cars to be in the streets?

The wonderful security mechanism of liability, lawsuits, and insurance.

This is interesting. So Google can claim that 160K mile of automated driving with the driver on-board and a single accident is blamed on the human.

How many times in that 160K miles were accidents prevented by vigilant human override?

The security mechanism _is_ the human driver, for now - the only reason they're legally allowed to run these cars is because there's a human standing by to take over control at a moment's notice.