Monitoring software flagged Black Lives Matter tweets.
Twitter has confirmed to The Daily Dot that it has terminated access to user data for Media Sonar, a Canadian company that marketed social-media monitoring tools to U.S. police departments. The service offered to help police “identify illegal activity and threats to public safety” by flagging keywords referencing drugs, bullying, and prostitution.
Get Data Sheet, Fortune’s technology newsletter.
But flagged terms also included “policebrutality,” “Dissent,” and “justiceformike,” a reference to the 2014 shooting of Michael Brown which sparked widespread protest. That fact was uncovered last year through a public records request by the ACLU of Northern California, which called social media monitoring “part of a pattern of unchecked surveillance” that “risks targeting communities that are already vulnerable to police misconduct.”
Media Sonar, and software like it, builds profiles of social media users based on their posts’ content, links, and context. The ACLU also highlighted a piece of surveillance software called Beware that assigned individuals “threat levels” of green, yellow, or red based on their social media activity. Such software also collects location data on users’ posts.
That behavior violates Twitter’s stated position on surveillance, particularly of activists and protesters. As spokesperson Chris Moody wrote last month, “our commitment to social justice is core to our mission,” and “we prohibit developers using the Public APIs . . . from allowing law enforcement—or any other entity—to use Twitter data for surveillance purposes. Period.”
For more on digital surveillance, watch our video.
Twitter told The Daily Dot that it had terminated Media Sonar’s data access in October. That followed similar actions against similar surveillance companies Geofeedia and Snaptrends. After losing its access to both Twitter and Facebook data, Geofeedia reportedly cut more than half its staff and chose to shift the company’s focus.