Future Robot Security officers and a future legal nightmare of liability

Security robot officer: "Mam. Can I please help you across the street?"
Old lady 80 years old:"Why yes, young man. I would appreciate that."
As robot security officer grabs the old lady's hand his grip is wrong and he breaks it in three places in a multiple compound fractures. The old lady screams in pain and faints. Then the robot security officer is programmed to take this lady to a hospital. So, he picks up the old lady but comes too close to a light post with her head because he isn't used to allowing for her dimensions in normal travel. The lady is concussed by this lamp post at the speed at which the robot is traveling and has an  brain aneurism and dies on the way to the hospital.

Now, who is liable in this situation? Is it the manufacturer of the robot? Likely no because they didn't contract for this kind of event. Is is the city who hired this robot as a police Officer? Maybe. Is it the lady herself for allowing this thing to hold her hand? Likely yes. But that doesn't really matter does it because she is now dead.

So now you can see just how awful the first security officer robots might be for the first 10 or 20 years of service until all the bugs are worked out of their software and hardware in relation to human beings. And even then strange things will happen as people die or are maimed in unexpected circumstances like when it rains and software shorts out in a robot security robot somewhere and problems arise or when regular maintenance isn't conducted in good order in good time, or when lightning strikes the robot who is holding a child's hand at the time and both the child and the main CPU of the robot go down at the same time.