At 11:27 on Monday morning, as the family of a Baltimore man who suffered a fatal injury while in police custody gathered at the New Shiloh Baptist church for his funeral, the Baltimore Police Department used its Twitter account to spread news of a “credible threat to law enforcement” from gangs, who had allegedly united to “take out” police officers. The police offered little corroborating evidence or information about where the tip came from, but the media took the story and ran. The narrative of the Baltimore police as victims spread quickly, and tinted much of the coverage of the protests later that night.

There had in fact been a report earlier Monday that the Bloods and the Crips and the Black Guerrilla Family had come together—but not for some sort of cop-killing strategy summit. Instead, they were showing respect for Freddie Gray at a demonstration at Baltimore’s City Hall. “We did not make that truce to harm cops,” one gang member told a television reporter. “To stop what’s going on—that’s all we’re trying to do. We want justice for Freddie Gray.” A Crips member identified as Charles told The New York Times that he and others tried to protect black-owned businesses from looters—though he admitted they directed vandals “toward Arab- and Chinese-owned stores” instead.

The “credible threat” alert was just one of many pieces of information spread by the Baltimore Police Department via its official Twitter account that seemed to cross the line between public-safety information and propaganda. The feed from Monday night sounded like a dispatch from an urban dystopia in which there are only a few good guys with badges, against everyone else. Many of the tweets are written in language that does more to evoke fear and lay blame than to inform. “In an act of violence and destruction—a group of criminals have set another car on fire at North Avenue and Fulton Ave,” read one tweet. Another: “Groups of violent criminals are continuing to throw rocks, bricks, and other items at police officers.” The word “criminals” appears over and over again, so often that someone reading the feed could be forgiven for thinking that Baltimore really did turn into a scene from The Purge.

Some of the information that the department has provided to its 127,000 Twitter followers seems to have been at best incorrect and at worst deliberately misleading. “A group of criminals have just started a fire outside the library located at Pennsylvania Ave and North Ave,” the police tweeted on Tuesday night. But according to Guardian reporter Jon Swaine, it was the police themselves who caused the fire, when sparks from a tear-gas grenade landed on trash.

Individual tweets were often framed as safety advisories, but together they created a selective narrative of events that later bled into news coverage. On Monday afternoon at 3:01 pm the department warned on Twitter and Facebook about “a group of juveniles in the area of Mondawmin Mall. Expect traffic delays in the area.” Half an hour later the police said that kids had started throwing bricks; 15 minutes after that the department reported that an officer had been hurt. Later, media accounts would describe a violent riot started by teens who were hungry for a fight. But as eyewitnesses pointed out, the cops had shown up in full riot gear just as a high school near the mall was letting out. The police shut down the subway station and the bus lines, effectively trapping the students. “Those kids were set up, they were treated like criminals before the first brick was thrown,” one teacher wrote. Social media functioned as a sort of virtual riot gear, manufacturing the narrative of violence in the digital realm as the police were escalating it on the ground.

“Someone follows the police Twitter feed for two reasons, the first being public safety. I want to know if there’s a crime in my neighborhood, or a road blockage,” said Jacob Simpson, the pastor of the Salem Lutheran Church in south Baltimore. “Or, someone follows it as a cheerleader of the police department.” Simpson’s impression is that the police are using social media to cater to the latter, and at the expense of safety. “Their job is to provide a basic account of where hotspots are, but they are editorializing…. They’re galvanizing people who have this nasty narrative about what’s going on.”

Lawrence Brown, a public health professor at Baltimore’s Morgan State University, sees the police department’s social media strategy as part of a broader campaign to darken public opinion towards demonstrators and build sympathy for the police. “Why would they tweet this out in the middle of a funeral?” he asked, referring to the “credible threat” announcement. Brown traced a pattern in the communications of city officials, from the police union president comparing protesters to a “lynch mob” to the governor, the mayor, and the city council president’s invocation of “thugs” to the police department’s repeated use of “criminals.”

“The protesters were winning the narrative in local media and so it seemed like [city officials] were trying to find a way to capture the dominate position,” said Brown. “Once you paint them as the enemy to the media, to the general public, then you can really justify the way you go about treating the people after that.”

The Baltimore Police Department is not the only law enforcement agency to use social media to shape a narrative rather than as a conduit for safety information.

In December, after five football players from the St. Louis Rams walked onto the field with their hands up in a show of solidarity with protesters in Ferguson, Missouri, the St. Louis Police Department used Twitter to declare that an expression of regret from an NFL executive was not good enough.

And police departments far from Baltimore have now joined that department’s social media PR campaign.

The Baltimore Police Department did not respond to an inquiry about management and oversight of its Twitter feed. The department’s activity on social media over the past several days has raised a number of bigger questions, too, questions that are likely to come up again as law-enforcement agencies increasingly find their narratives challenged by skeptical citizens. What is the purpose of law-enforcement agencies’ social-media communications—is it appropriate to use them as tools for publicity and commentary, as well as to distribute public safety information? How can that line be meaningfully drawn? Does using language like “criminals” to dehumanize whole groups of people—who have not been charged with a crime—really promote peace in a community, or instead make it harder to achieve?