zlacker

It is definitely different. I do think they could put in a request like "give me all footage between these hours in this area", and then do the facial recognition themselves.

It's conceptually pretty similar to cell tower dumps, where they ask for all data from a cell tower during a particular time frame. This was recently ruled unconstitutional (https://www.courtwatch.news/p/judge-rules-blanket-search-of-...), but they used it for like 15 years before that (https://arstechnica.com/tech-policy/2013/08/how-cell-tower-d...). I can imagine blanket car footage dumps working for a similar amount of time.