Want the best of VICE News straight to your inbox? Sign up here.Gettr, the pro-Trump Twitter alternative launched last month by close Trump adviser Jason Miller, is allowing users to share child exploitation images.New research from the Stanford Internet Observatory’s Cyber Policy Center has laid bare the dangers of the platform’s almost complete lack of moderation and identified more than a dozen child abuse images being shared by Gettr users.
Unraveling viral disinformation and explaining where it came from, the harm it's causing, and what we should do about it.
“Social media platforms typically use machine-learning models to analyze uploaded image and video content to determine how to act on it—uploads can be rejected entirely, put behind a sensitive content filter, or, in severe cases, reported to law enforcement,” the researchers wrote.One of the most widely used pieces of software is known as PhotoDNA, which stores a list of child exploitation imagery, against which platforms can check uploaded content before it is allowed on the site. But Gettr does not use PhotoDNA or any other automated software to monitor the content being uploaded to its site. Instead, it relies on users to report content.“This, frankly, is just reckless,” David Thiel, an author of the report, tweeted. “You cannot run a social media site, particularly one targeted to include content forbidden from mainstream platforms, solely with voluntary flagging. Implementing PhotoDNA to prevent CEI is the bare minimum for a site allowing image uploads.”By using PhotoDNA’s database of images, the Stanford researchers were able to identify 16 matches among a sample of images taken from posts and comments on Gettr. They were also able to successfully show how easy it is to upload child exploitation imagery by posting several benign images PhotoDNA stores in its database for testing purposes.Miller told VICE News that Stanford’s report is “completely wrong,” claiming that Gettr has “a robust and proactive, dual-layered moderation policy using both artificial intelligence and human review, ensuring that our platform remains safe for all users.”
Miller also claimed that all images of children uploaded to the site are reviewed by a human moderator. However he did not immediately respond when asked about the 16 matches for child exploitation images in the PhotoDNA database that the researchers found.The Gettr app is available on both Apple and Google’s official app stores, but neither company responded to VICE News’ request for comment about whether the sharing of child abuse imagery on Gettr breached their terms of service. Gettr was launched on July 1 as an “non-bias” alternative to Twitter, where users could post content more freely. The site initially banned curse words, however, and its terms of service forbid a very wide range of content, including posts that are “unlawful, offensive, obscene, lewd, lascivious, filthy, pornographic, violent, threatening, harassing, or abusive, or that violate any right of any third party.”Miller initially boasted about Gettr’s explosive growth, claiming in the middle of last month that the site had registered 1.5 million users in just 11 days. But the researchers found that the Gettr support account, which is automatically followed by all new users, didn't reach the 1.5 million mark until the first week of August. Further analysis shows that new-user registrations have dramatically slowed to a trickle, just over a month after the site launched.
And the vast majority of the people who have registered have yet to post or comment on the platform. Of the approximately 1.5 million users with accounts, just 55,000 of those have posted 10 or more times. There are also a considerable number of spam accounts among Gettr’s registered users.
This is hardly surprising however, given that in the week after Gettr launched, the site’s source code was leaked and prominent accounts were defaced. The platform was also spammed with Sonic porn posted by left-wing trolls, who pointed out Miller’s hypocrisy when he banned the content on the platform.Then, two weeks ago, a Politico investigation found that the site was filled with extremist jihadi content, including beheadings and viral memes in support of the Islamic State.Stanford’s research found that the site was filled primarily with far-right users from the U.S. and Brazil who were deplatformed by larger social media sites, as well as a sizable Arabic-speaking population.
Among the most popular accounts are former Secretary of State Mike Pompeo and right-wing commentator Dinesh D’Souza. The accounts belonging to right-wing media outlet Newsmax and Steve Bannon’s “War Room” show are also very popular.Pompeo’s account claims he has 1.6 million followers on Gettr, even though there doesn’t appear to be that many people registered on the platform. This is likely a result of the Gettr feature that allows new users to import their followers from Twitter, even though the majority of those users are not on Gettr.
For Brazilian users, the verified accounts of authoritarian President Jair Bolsonaro and his son Flavio Bolsonaro are the most popular.Another indication of Gettr’s complete inability to moderate content on its site was a successful spam campaign has led to #transrights and #transrightsarehumanrights appearing in the top 10 trending hashtags on the platform. This was almost entirely the result of one single user posting the same comment repeatedly, something that all major social networks would have quickly and easily detected and shut down.