It began with the 1999 mass shooting at Columbine High School in suburban Denver, when two students opened fire and killed 15 people, including themselves.
The aftermath of the tragedy saw the debut of Safe2Tell, a tool that lets parents and students anonymously report information about potential school shootings and other violent threats. Safe2Tell is still in use today. But it began with a simple data point: In most school shootings, at least one person had information about the plans prior to the attack.
“People will call and report, ‘My friend said that they were going to kill themselves,’ and then law enforcement will go out and investigate, and in many of these situations, lives have been saved,” said Beverly Kingston, director of the Center for the Study and Prevention of Violence at the University of Colorado Boulder and a former Safe2Tell board member. “So it’s one of these things [that poses the question], ‘Why don’t we have this system throughout the nation to do this?’ It just helps protect everyone’s safety.”
Nearly two decades after Columbine, such tools have become an industry unto themselves and extend far beyond anonymous reporting. Particularly in the aftermath of this year’s shootings in Parkland, Florida, and Sante Fe, Texas, school and law enforcement officials are turning to social media as a way to head off potential violence before a single shot is fired. The reason: As a key platform for student self-expression, social media has also become an outlet for youth to broadcast school shooting threats and suicide warnings.
A Toy Gun, a Snapchat Post, and an Arrest
New tools use artificial intelligence to search for potential threats on students’ social media profiles and scan school-issued laptops in search of keywords that could spell trouble. After the Parkland shooting, lawmakers in Florida took it a step further and mandated a new database that combines law enforcement and social services records with social media activity to help officials investigate students who post suspicious or threatening information online.
While the tools promise to protect kids, they also pose questions about how far schools can encroach on student privacy in the name of keeping them safe.
Every bit of information could be helpful during emergency situations, said Kenneth Trump, president of National School Safety and Security Services. But he questioned the extent to which schools should be asked to monitor students’ off-campus behavior.
“The question becomes, ‘Where do you draw the line?’” he said. “How much can we reasonably expect of a building principal or a system principal or maybe a dean to be the internet police?”
‘Outcries for help’
To Gary Margolis, helping school officials identify online threats is crucial. Formerly the police chief at the University of Vermont, he observed the growth of social media from a law enforcement perspective. In the internet age, he said, “outcries for help” moved beyond students passing notes or exchanging whispers in the locker room.
Now, those outcries occur online.
In response, Margolis founded Social Sentinel to help schools pick up on troubling social media posts. Social Sentinel collects social media data and uses artificial intelligence to run posts against a “library of harm” containing some 450,000 phrases, keywords, hashtags — even emojis — that he said could indicate a suspicious post. School districts are then notified, via email or text message.
Inside the $3 Billion School Security Industry: Companies Market Sophisticated Technology to ‘Harden’ Campuses, but Will It Make Us Safe?
“In the absence of a service like Social Sentinel, it’s impossible to be part of the conversation,” he said. Once the program flags content for school officials, it’s up to them to act on the information. He declined to provide specific examples of how the platform has been used to thwart violence.
Among the districts using the tool is Ohio’s Lebanon City Public Schools. Each morning, Superintendent Todd Yohey, school police and the district human resources director get posts the platform flags as potentially problematic. The tool doesn’t offer identifiable information beyond the social media accounts the post came from, but Yohey said anonymity isn’t a big hurdle.
“Kids aren’t very good at hiding their identity,” he said. “There are times we’ve asked other kids, ‘Hey, do you know who this poster is?’” since students generally know their peers’ social media accounts.
Social Sentinel generally flags about one social media post a day for school officials in Lebanon City, Yohey said, and more often than not, the posts are benign. During the NBA playoffs last year, for example, the tool flagged multiple posts.
“There were a lot of posts about, ‘That three-pointer was the bomb,’ and so the word ‘bomb’ would trigger a notice to us,” he said. “You read that and you figure out, ‘Oh, they’re talking about the basketball game, and so it’s not concerning.’”
Still, stumbling onto just one threatening post would make the effort worthwhile, Yohey said, if it helped the school avert a tragedy. The tool would allow the district to react to a situation more quickly. “If we get a post that said, ‘I’m planning to take a bomb to school tomorrow,’ or ‘I planted a bomb at the elementary school,’ those of course would require immediate attention.”
In fact, a day after the Parkland shooting, a district student was arrested for texting a friend that he was going to shoot up a school. In response, Yohey issued a warning to students: “There is no such thing anymore as an empty threat; no jokes, no kidding around, no ‘I didn’t mean it.’”
Although mass shootings this year caused a spike in anxiety and prompted heated policy debates, they are statistically rare and campuses have actually become safer in recent years, according to recent National Center for Education Statistics data.
Chad Marlow, senior advocacy and policy counsel at the American Civil Liberties Union, said he’s concerned about “subtle harms” schools could inflict by surveilling students. Monitoring could signal to students that they can’t use social media freely to share information with their friends. As schools crack down on student speech, students of color — who already face disproportionate suspension rates — could be most affected.
“The rules governing these things involve standards that are uncomfortably subjective,” he said. “Monitoring places many groups of students, who are already vulnerable by virtue of the communities that they are members of, in even greater places of vulnerability.”
In one instance, in 2013, an Alabama school district hired a security consultant to monitor the social media accounts of hundreds of students — an effort that the superintendent claimed, somewhat dubiously, was launched after the National Security Agency alerted school leaders to a student’s social media threat. The district surveillance effort resulted in the expulsion of 14 students, 12 of whom were black. While some students were expelled for posting pictures with guns, others were disciplined for “holding too much money” and for making the “OK” sign, which police said was a gang symbol.
In interviews, several social media monitoring companies have deflected concerns about government surveillance by noting that the services flag only content posted publicly on sites like Twitter and Facebook. Students concerned about privacy, company executives said, can update their privacy settings so their posts are only accessible to specific people.
Bowing to concerns from civil rights groups, an executive at one social media monitoring company said he changed his platform. The executive, Soter Technologies founder and CEO Derek Peterson, said the company disabled a feature in social media tool Digital Fly that allowed school districts to create a list of students with a history of posting troubling information online.
But the ACLU’s Marlow argues that social media monitoring programs meet the “textbook definition” of mass surveillance because they watch a large number of people at the same time.
“They can call it rose pedals in a field of daisies if they want, but it is absolutely surveillance,” Marlow said. “And surveillance has chilling effects on free speech.”
Monitoring students online doesn’t stop with social media. As more schools adopt “one-to-one computing” programs that place laptops in the hands of each student, some districts have equipped them with monitoring programsthat scan online activity like emails and search histories for signs of threats or illicit activity such as drug use. These tools have prompted concernsfrom digital privacy groups like the Electronic Frontier Foundation, which noted in a 2017 report that students are “backed into a corner” because they have no choice but to use the school-issued devices.
In one instance, in 2010, a Pennsylvania school district paid more than $600,000 to settle an invasion of privacy lawsuit after a high school student accused it of using the webcams on district laptops to take covert photos of students — including a picture of the teenager sleeping in his bed. The district later acknowledged it had taken more than 50,000 such pictures in an attempt to recover lost or stolen devices.
Karen Gullo, the foundation’s spokeswoman, said in a statement that monitoring tools that flag specific words “can cause false alarms and target innocent students.”
“Pervasive surveillance normalizes electronic snooping,” Gullo said, “and can keep kids from testing out new ideas and identities as they grow.”
‘See something, say something’
Ironically, after a decade of online innovations, some observers say the most effective, and least invasive, method of neutralizing threats remains the low-tech idea that surfaced nearly 20 years ago: tips from actual people.
Schools across the country have adopted the “See something, say something” mantra popularized after the Sept. 11, 2001, terrorist attacks. A half-dozen states already have anonymous tip lines replicating the Colorado model, and officials from more than 20 states have contacted Safe2Tell to inquire about the program since the Parkland shooting, a spokeswoman for Colorado’s attorney general said.
In October, the Justice Department announced more than $19 million in grants to develop anonymous reporting systems and threat assessment teams for schools. Congress appropriated the funds last spring through the STOP School Violence Act. Sandy Hook Promise, a nonprofit that operates the Say Something anonymous reporting system, lobbied for the federal law. School districts in nearly half the states in the country are implementing the Say Something system.
Last school year, Colorado’s reporting tool collected 16,000 anonymous tips, a 74 percent surge from just one year earlier. Post-Parkland, anonymous tips surged even more, Colorado attorney general Cynthia Coffman said in an interview.
“Kids were reacting to everything they saw and heard, particularly on social media,” Coffman said. “We’d have multiple reports from kids about the same social media post that said something about Parkland.”
Despite the increase in anonymous tips, it remains unclear what percentage of them represented true threats and resulted in intervention. Safe2Tell is collecting that data now, Coffman said. Still, the program boasts that school and law enforcement officials have intervened in multiple suicide and school shooting threats based on the tips.
But even this low-tech, relatively noninvasive approach has drawbacks. On several occasions, people have used Safe2Tell to bully other students by filing false reports. In one case, a 15-year-old girl claimed someone filed three false tips that she was suicidal and used drugs. On two occasions, she said police pulled her out of class to question her.
In order to discourage false reports, Coffman said people can be charged with a misdemeanor if they use the system to harass others.
While the bevy of tools on the market for educators to see students’ online activity are helpful, officials need to put the information into context and consider the student’s broader behavior, said Marisa Randazzo, managing partner at SIGMA Threat Management Associates. When confronting students about online posts, they should avoid a zero-tolerance response like automatic suspensions, she said.
“That’s a policy that sounds good but actually has an inadvertent chilling effect on students’ willingness to come forward and share information and share what’s worrying them,” she said.
Kingston, of the University of Colorado Boulder, offered similar advice, noting that school and police officials should respond to perceived student threats with caution.
“Adults who are entrusted with these jobs have to walk this delicate rope of also making sure they’re protecting the safety of everyone,” she said. “We want to make sure that our interventions don’t do more harm to students.”