Every social media network has its underbelly, and the one on Periscope, Twitter’s live-video app, might be uglier than most: On any given day, users appear to flock to broadcasts from minors and encourage them to engage in sexual and inappropriate behavior. Worried Periscope users have been ringing the alarm for more than a year, and Twitter has reaffirmed its zero-tolerance policy against child exploitation after reporters have followed up. But if the company has been working any harder to enforce that policy, its efforts don’t appear to have scrubbed out the grime.
Last month, a tipster described to me how some Periscope users were routinely pursuing children who had logged on to the platform to play games like truth or dare with others. It took pseudonym-cloaked commenters less than six minutes to persuade a girl, broadcasting with a friend and playing truth or dare on a public forum recently, to lift her shirt and show her breast. “Fully out,” typed one user, right before the girl revealed herself. “But with shirt up…” instructed another, before the girl did it again. The girls, both of whom had braces and appeared to be younger than 18, said they loved to roller-skate, mentioned their homeroom class, and said they didn’t know what an “underboob” was after being asked to show some. It’s not clear whether the users directing the girls were also minors or were adults. But whatever the age of the commenters, their behavior was in violation of Periscope’s standards, which bars users from engaging in sexual acts and “directing inappropriate comments to minors in a broadcast.”
In another alarming video, a pair of girls who described themselves as sisters (one said she was 14, and the other appeared to be several years younger) were asked to show their bras and their underwear and pressured by multiple commenters to continue to strip. “Dare y’all to play rock, paper, scissors, and loser has to flash,” said one viewer, after both girls had already shown their underwear.
Launched in 2015, Periscope makes it easy for anyone to start a broadcast that others can watch live and send comments to the broadcaster while he or she is filming. Commenters can also send broadcasters hearts to show that they’re enjoying the live content. As you Periscope, you can see the comments and hearts in response to your stream. There is also a private stream function, which is only available to users who follow each other. In incidents like the ones described above, commenters routinely ask the young broadcaster to follow them, perhaps hoping to engage in a private video stream.
Although concerned Periscope users have been alerting the company that some people were using its app to coax children into inappropriate behavior for more than a year—and in July, the BBC even aired an investigation into how users on Periscope were pressuring children with sexually explicit messages—children and teenagers can still be swamped with requests from viewers to do things like take off their shirts and pants, show their underwear, show their feet, kiss other kids, do handstands, and answer lewd questions. In other words, it’s clear the company hasn’t figured out how to solve the problem. In response to the BBC’s reporting, Periscope said, “We have a strong content moderation policy and encourage viewers to report comments they feel are abusive. We have zero tolerance for any form of child sexual exploitation.”
It’s not that Periscope hasn’t done anything. On Nov. 27, about five months after the BBC report, Periscope rolled out an update to its reporting tool that allows users to flag potentially inappropriate content. The updated tool includes a category for “child safety,” as well as a way to flag “sexually inappropriate” comments by users talking to broadcasters on livestreams. In that announcement, Periscope said that since “the beginning of 2017, we have banned more than 36,000 accounts for engaging or attempting to engage inappropriately with minors.” This announcement, however, came in the form of a post on Medium (where Periscope only has 116 followers), which the company tweeted out five days after publishing it, after updating it to add details on the new reporting tools. In the app itself, there was no announcement or indication that the new feature existed that I’ve been able to find, suggesting that many Periscope users might be unaware of the updated reporting tool.
I contacted Periscope on Nov. 30 to ask about explicit interactions with minors on the platform and what the company is doing to solve the problem. In response, Periscope encouraged me to report any problematic videos found in the future and said that it has “a team that reviews each and every report and works as quickly as possible to remove content that violates our Community Guidelines.” I then asked about the size of the team, which Periscope said in its recent Medium post is expanding, and asked for more information about what else the company is doing about this kind of content. I haven’t heard back but will update this piece if I do. I also asked the Department of Justice if it was aware of and had taken any actions regarding this activity on Periscope. A spokeswoman said, “As a matter of policy, the U.S. Department of Justice generally neither confirms nor denies the existence of an investigation.”
In its Medium post Periscope did say that it’s “working to implement new technology” that is supposed to help detect accounts that are potentially violating the company’s policy and improve the reporting process—though at the moment, it’s not clear whether that software is running or the company is relying on user reporting alone. (When pressed on that question, Periscope did not respond.) Due to the live nature of the videos, it’s probably hard for Periscope to know exactly when a new one pops up that features a minor and attracts predatory commenters, though the platform has removed live broadcasts while they are happening in the past. “Unless they’ve got keywords down really tightly to know what constitutes a grooming message, … automated detection may be a little harder to do just via existing algorithmic tools,” Thomas Holt, a criminal justice professor at Michigan State University who specializes in computer crimes, told me. That means that having a reporting feature to help target accounts for removal is critically important, as is having staff to review the user reports. But, according to Holt, the efficacy of those reporting tools depends on how much users are even aware they exist. Kids might not even know when a pedophile is attempting to lure them into sexual acts, or even that it’s wrong and should be reported. And again, even a strong reporting regime clearly isn’t enough.
Videos of children being lured into sexual or inappropriate behavior on Periscope can rack up more than 1,000 views. The videos tend to follow a pattern: Once the stream starts, dozens of Periscope users flock into the comments, as if they had been alerted either on Periscope or via a separate forum outside of Periscope, suggesting some level of coordination. This type of swarming is common, according to Holt: “Multiple people will often start to send sexual requests, questions, or content in an attempt to exert a degree of social pressure on the person to respond to a request.” This makes the request seem more normal, Holt says, and can manipulate a child to respond to a sexual request to please the group.
One place within Periscope that had become a hive for this kind of misbehavior was the “First Scope” channel, which curated streams from people using the platform for the first time, according to Geoff Golberg, a former active Periscope user who has been vocal in calling attention to the problem of inappropriate behavior directed toward minors on the app. That channel was removed in November, months after Golberg sent emails to the company (which he tweeted out) about the potential of minors being sexually exploited in the channel.*
While it’s good that Periscope is taking some degree of action, Holt says that the risk posed by virtually every social media platform—particularly ones that are more reliant on images than text, since text is easier to patrol with software—means it’s critically important for parents to understand what their kids are doing when they’re online, and to have conversations with them about what apps they use, what constitutes bad behavior, and how to report it. Periscope isn’t the only popular social media site struggling to moderate how kids use the app. Last month, the New York Times reported how the YouTube Kids app hosted and recommended videos with disturbing animations of characters killing themselves and committing other violent acts. On Periscope, though, the dangers are heightened because of the live, instant nature of the broadcasts, which can put a mob of predators in conversation with children before there’s time to intervene.
In many ways Periscope is a remarkable service, allowing anyone to share what they’re doing in real time with viewers around the world, whether it’s a confrontation with law enforcement or a hot-air balloon ride. But it also facilitates behavior that calls into question the utility of the entire enterprise—and how capable the company is of curbing that behavior effectively, either through moderation or software. Over at Alphabet, YouTube is attempting to fix the problems on YouTube Kids by hiring more moderators. Twitter and Periscope should do even more than that. The safety of some of its most vulnerable users is at stake.
*Correction, Dec. 18, 2017: This article originally misspelled Geoff Golberg's last name. (Return.)