The controversy that swept up leading tech companies last year over whether facial recognition software is accurate enough to be sold to law enforcement agencies shows no signs of abating. Last week, a group of Amazon shareholders demanded that it stop selling the technology to police, and pushed for a vote on the matter at the company’s shareholder meeting in May.

Now, a growing number of startups that sell facial recognition technology are also getting embroiled in the ethical implications of the technology, which civil liberties groups and activists warn is inaccurate and can worsen discrimination in policing. In all-hands meetings over the past few months, employees at Clarifai, an artificial intelligence startup whose software can recognize the content of images and video, complained to CEO Matthew Zeiler about the startup’s plan to launch an as-yet-unannounced facial recognition product.