The third of these is exactly what I'm looking for: a case where a Flock camera was off by one letter, and mismatched a car as stolen when it wasn't. The driver wasn't arrested, but being stopped for a stolen vehicle is terrifying --- it happened several times where I live as well (due to stale Illinois LEADS entries).

The first two don't seem like Flock issues at all.

In the first, police used Flock cameras as they would have a closed-circuit camera. They got a vague look at the person driving a suspect car, and they got another lead on the owner of the car, tracked him down on Facebook, and picked out a person from his friends list that looked like the picture on the camera.

In the second, there was a fatal hit-and-run accident involving a Dodge Durango; the police did a dragnet search for Dodges Durango and literally picked the first one they could find parked on the street. They then papered the case backwards to match it to the accident, claiming it had telltale damage (it did not).

Both those cases are police malfeasance, to be sure. But they're not the police believing Flock against all other evidence; they're police essentially manufacturing evidence to clear a case.