In the News: Manjeet Rege on Limits of Facial Recognition Technology - Newsroom | University of St. Thomas

March 26, 2026 - 01:00

The growing use of facial recognition technology by law enforcement carries significant risks of misidentification and wrongful accusations, according to a leading data science expert. Manjeet Rege highlighted the critical limitations of these systems, emphasizing that their improper application can derail criminal investigations and harm innocent individuals.

Rege explained that many algorithms are trained on datasets lacking demographic diversity, leading to higher error rates for people of color, women, and younger populations. This inherent bias means the technology is not equally reliable across all community groups. Furthermore, using low-quality images from surveillance cameras or social media, often grainy or taken at poor angles, drastically increases the chance of a false match.

He stressed that a facial recognition "match" should never be considered definitive proof, but rather a single piece of potential leads that requires traditional, corroborating evidence. Relying on it as a sole identifier, Rege argues, is a dangerous shortcut that undermines justice. The call is for stringent oversight, transparent policies on use, and a clear understanding that the technology is an imperfect tool, not an infallible solution. As adoption spreads, balancing innovation with ethical safeguards and accountability remains paramount.