In early tests, the software identified proper mask use from security cameras with 80 percent accuracy. While it might not be used to find individual infractions, the system could spot “hot zones” for mask misuse, the company’s founder Jaz Banga told Bloomberg. Teams are reportedly considering non-threatening enforcement methods, like sending out someone in a nurse’s uniform to hand out free masks.

Facial recognition is already a controversial subject over privacy concerns, with firms like Clearview under heavy regulatory scrutiny. Using it as part of a strategy to bring back fans could rile critics, considering that the league hasn’t been able to control the virus among its own players. And even gentle enforcement might be problematic, given that mask-wearing has become such a political issue.

Banga emphasized that the system would use pixelization to make faces unrecognizable and said “we don’t even know how to” recognize faces. In any case, it would probably take a while to approve and implement the system, so the cardboard and virtual fans are likely here to stay for the foreseeable future.