> That mechanism is judicially auditable. The AI is not. We should not be arresting people based on the output of an unauditable mechanism.

Are we, though? If “ClearView AI gave us a hit” is being treated as probable cause sufficient for an arrest warrant, that's a problem.

If it's being used as a tool to generate leads to investigate and traditional evidence is gathered and presented, I don't see a big problem.

It’s not auditable and scale can make it a substantially different problem altogether. But, if you’re in a restricted area or a public building I think it makes sense.