To be honest, when I wrote that I was imagining a step that probably doesn't exist yet, which would be connecting the facial recognition to some sort of LMM tha

To be honest, when I wrote that I was imagining a step that probably doesn't exist yet, which would be connecting the facial recognition to some sort of LMM that would search the web along with private databases and filter results based on context before showing them to a human. So if, e.g. a photo of a person popped up on a camera in Las Vegas an hour after a murder committed by a very similar-looking person in Miami, the Vegas photo would be ruled out. Such an engine would be prone to all kinds of spoofing. I wasn't really thinking about the competence or vigilance of the police (which I think, as in any profession, varies wildly from place to place, officer to officer, and one circumstance to another). Even as it stands, though, the technology seems vulnerable to deepfakes. Someone in control of a private camera that fed into the database could get away with murder and frame someone else for it.