Twitter accounts offering to trade or sell child sexual abuse material under thinly veiled conditions and hashtags went viral for months after CEO Elon Musk said he would fight child exploitation on his platform. Stay online.
“Priority 1‘ Musk said in a tweet on Nov. 20. He also criticized former Twitter leaders, claiming they did little to address child sexual exploitation and intended to change the situation.
But since that declaration, at least dozens of accounts have continued to post hundreds of tweets in total, using terms, abbreviations and hashtags to denote sales. What Twitter calls child sexual exploitation material, according to the count of tweets for just one day. Signs and signals are well known among experts and law enforcement agencies working to stop the spread of such material.
Tweets reviewed by NBC News are intended to sell or trade content commonly known as child pornography or child sexual abuse material (CSAM). The tweet did not show CSAM, and NBC News did not show it in the process of reporting this article.
Some tweets and accounts predate the Musk takeover by months. They remained live on the platform as of Friday morning.
More tweets reviewed by NBC News over the course of several weeks were published during Musk’s tenure. Some users tweeting the CSAM offer apparently deleted the tweet shortly after posting to avoid detection, then posted similar offers from the same account. Some accounts offering CSAM said he was able to create a new account while his previous account was closed by Twitter.
according to twitter rules Published in October 2020, “Twitter will not tolerate any material that features or promotes the sexual exploitation of children, which is one of the most serious violations of the Twitter Rules. may contain media, text, illustrations, or computer-generated images.”
In an email to NBC News after this article was published, Ella Irwin, Twitter’s vice president of product overseeing trust and safety, said: We’ve found a lot more than Twitter has seen in a long time, but we’re doing a lot to keep improving. Irwin asked that NBC News provide its findings to the company so it could “follow up and withdraw the content.”
It’s unclear how many people remain on Twitter to deal with CSAM after Musk enacted several layoffs and issued an ultimatum that led to a wave of resignations. Musk has outside help, the company said in December A sharp increase in child sexual exploitation account suspensions. A representative for the Center for Missing and Exploited Children, a U.S. child exploitation watchdog, said the number of her CSAM reports detected and flagged by the company hasn’t changed since Musk’s acquisition.
Twitter has also dissolved its Trust and Safety Council, which includes a nonprofit focused on responding to CSAM.
The company employed more than 7,500 people as of the end of 2021, according to Twitter’s annual report to the U.S. Securities and Exchange Commission. According to internal records obtained by NBC News, Twitter’s overall employee count had fallen to around 1,340 active employees as of early January, down from around 20. People who work in his Trust & Safety organization at the company. That’s less than half the number of former Trust and Safety employees.
A former employee who worked on child safety issues was a specialty of a larger trust and safety group, and was on a team that implemented anti-CSAM rules and related violations prior to Musk’s acquisition of many products. He said the manager and the engineer had left the company. The employee requested anonymity because he signed a confidentiality agreement. We don’t know the exact number of people Musk assigned to these tasks.
Twitter has halved the number of engineers at the company since Musk took over the platform, according to internal records and people familiar with the situation.
“Many employees who were on the child safety team last year are no longer part of the company, but Twitter has seen a rapid increase in personnel across the company, mainly between January and August of last year,” Irwin said in an email. She added that the company “has about 25% more staff on this issue/problem area now than it did at its peak last January.”
CSAM has been a perennial problem for social media platforms. Also, some techniques have been developed to automate the detection and removal of CSAM and related content, but as the problem evolves and changes, human intervention remains necessary and the UK’s child exploitation crimes continue to grow. expert Victoria Baynes said. National Crime Agency, Europol, European Cybercrime Center, and Facebook.
“If you fire most of the human trust and safety staff who understand this, and rely entirely on algorithms and automated detection and reporting measures, you are just scratching the surface of the CSAM phenomenon on Twitter. ” said Baines. “We really, really need these humans to pick up signals of things that don’t look or sound right.”
The accounts NBC News confirmed promoting CSAM sales followed a known pattern. NBC News found that a tweet promoting CSAM’s deal posted in October was still alive — apparently undetected on Twitter — and the hashtag suggested users connect to other internet platforms. It has become a rallying point for providing information on how to trade, buy and sell. Sell exploitative material.
In a tweet seen by NBC News, a user claiming to be selling CSAM was able to evade moderation using easily decipherable, thinly veiled terms, hashtags, and codes.
Some of the tweets were brazen and their intentions were clearly identifiable (NBC News does not publish details about these tweets or hashtags, so they do not extend their reach further). The common abbreviation “CP,” short for “child pornography,” widely used online is not searchable on Twitter, but one user who posted 20 tweets promoting his material noted another searchable Using the hashtag, I wrote, “All CP collections for sale.” The tweet he published on December 28th. The tweet remained for a week before it appeared that the account was suspended after NBC News reached out to Twitter. A search on Friday found similar tweets still on the platform. Others used child-related keywords and replaced certain characters with punctuation marks such as asterisks, instructing users to send her messages directly to their accounts. Some accounts included the price in their account bios and tweets.
None of the accounts investigated by NBC News posted explicit or nudity photos or videos of abuse on Twitter, although some described images of young people dressed or undressed as “leaked.” ‘ or ‘decoy’ images with messages offering to sell them.
Many of the accounts using Twitter to promote harmful content point to the use of virtual storage accounts on MEGA, a New Zealand-based encrypted file-sharing site. The account posted a video of himself scrolling through MEGA, displaying folder names suggestive of child abuse and incest.
In a statement, MEGA Executive Chairman Stephen Hall said the company has a “zero tolerance” policy for CSAM on service. “If any public link is reported to contain CSAM, we will immediately disable the link, permanently close the user’s account, and provide details to the New Zealand authorities and relevant international authorities.” Mr Hall said. “We encourage other platforms to provide us with signals they perceive so that we can address Mega. Similarly, we will provide others with the information we receive.”
CSAM issues related to MEGA and Twitter have sparked at least one prosecution in the United States.
A June 2022 Justice Department press release found the sentence of an individual convicted of “forwarding and possessing thousands of images depicting child sexual abuse.”
“In late 2019, as part of an ongoing investigation, officers identified a Twitter user who submitted two MEGA links to child pornography,” the press release said. He admitted to viewing porn online and provided investigators with his MEGA account information, which was later found to contain thousands of files containing child pornography.”
Nearly all of the tweets seen by NBC News that promoted or promoted CSAM used hashtags that mentioned MEGA or another similar service, allowing users to search and find their tweets. . Hashtags are valid for months and are searchable on the platform.
This issue is widespread enough to attract the attention of some Twitter users. In 25 tweets, users tagged Musk using at least one of the major hashtags and warned him of the content. The first tweet, which flagged the hashtag to Mask with his username, said: [this] The hashtag has quite a few accounts requesting/selling CP. I was going to report them all, but there were too many and I got many more replies. Just keep your head up.
Historically, Twitter has taken action against some similar hashtags. For example, one of his hashtags related to his Dropbox, a cloud storage service, seems to be restricted in searches on Twitter right now. In a statement, a Dropbox representative said, “Child sexual exploitation and abuse does not exist at Dropbox and is in violation of our Terms of Service and Terms of Use. We use a variety of tools, including reviews, to find potentially violating content and take appropriate action.”
The automated systems used by many social media platforms were originally created to detect images and prevent their continued distribution online.
Facebook has been using a technology called PhotoDNA with human content moderators for 10 years to detect and prevent CSAM distribution.
Various companies have developed automated technologies to scan and detect text that may be related to CSAM. WhatsApp, a company owned by Meta, says it uses machine learning to scan the text of new profiles and groups in such languages.
A former Twitter employee said the company is working to improve its automated technology for blocking problematic hashtags. However, they stressed that human input would be required to flag and apply new hashtags.
“Once we know the hashtags we’re looking for, we automate the process of discovering hashtags for moderation. Human input is required to identify hashtags that may violate our policies.” ’” they told NBC News. “Today, machines generally automatically detect if a previously unseen hashtag is connected or may be in use by people looking for or sharing CSAM. It’s not taught to guess, it’s possible, but it’s usually quicker to use expert input and add it to the detection tool rather than waiting for the model to learn it Hashtag that is being used.”
In an email to NBC News, Irwin confirmed that “hashtag blocking was introduced a few weeks ago,” and said it needed human moderation. If we feel there is, we plan to automate it,” she added.
Equally important, according to Baines and former employees, is the fact that text-based detection can be overcorrected or pose a free speech problem. For example, MEGA is used for many types of content besides his CSAM, so the question of how to manage hashtags referring to services is not trivial.
“We need humans. That’s the simple answer,” Baines said. “And I don’t know if there are any people left doing this job.”
