AI IN CRIMINAL CASES IN 2025: USE OF FACIAL RECOGNITION TECHNOLOGY

Written by Hon. Abraham C. Meltzer

It would take a book to comprehensively discuss how AI is likely to impact criminal cases in 2025, so this article will focus on two cases that have received recent attention regarding the use of facial recognition technology (FRT) in criminal investigations.

Recent developments in FRT indicate that caution is needed. Live facial recognition has been implemented in the United Kingdom to search for wanted persons in real time in public spaces. But in the United States, facial recognition cannot yet, by itself, satisfy the standard of probable cause to arrest. A significant hurdle to using live facial recognition is the problem of detecting low base-rate events. Rite Aid Corporation’s foray into FRT provides a cautionary lesson on how live facial recognition can go astray. After Rite Aid’s failed experiment with installing FRT in its stores to catch shoplifters, the Federal Trade Commission sued it.and Rite Aid agreed to a five-year ban on using any facial recognition systems.

Nevertheless, FRT has already inserted itself into criminal cases, typically as the first link in a chain identifying a defendant. This raises the question of whether a defendant is entitled to discover the other matches a facial recognition system returned when it also matched to the defendant. The first court to address that issue, in Lynch v. State of Florida (2018) 260 So.3d 1166, held that the alternative matches are not discoverable. In contrast, in State v. Arteaga (2023) 476 N.J.Super. 36, a New Jersey court held that a defendant is not only entitled to discover the alternative matches, but also the source code and error rates of the facial recognition system used to identify the defendant. This question is likely to be revisited by other courts in 2025.