A shocking article was promoted on the “front page” or main feed of Elon Musk's X on Thursday.
The headline was “Iran attacks Tel Aviv with large missile.''
Captured X trends on April 4, 2024.
Credit: Screenshot from Mashable
This would certainly be a worrying development in world news.Earlier that week, Israel carried out An airstrike on the Iranian embassy in Syria killed two generals and other officers. Retaliation from Iran seemed a likely event.
Elon Musk's X was forcibly given a blue check. The situation is as follows.
But there was one big problem. Iran did not attack Israel. The headline was fake.
Even more concerning is that the fake headline was apparently generated by Grok, X's own official AI chatbot, and then promoted by Explore, X's trending news product, on the first day of the updated version of the feature.
Tweet may have been deleted
How did Mr. Musk make this happen?
A few years before Elon Musk acquired the platform then known as Twitter, the company rolled out new features, including: written context On trending topics. Technology to enhance trend description?Humans.
Previously, when a story broke, Twitter's algorithm caught related keywords posted and shared the trend, but it was sometimes unclear why a particular set of words was trending. .
However, in 2020, Twitter assembled the following team: human editor We curate news and put trends into context. In addition, the top tweets that appear under trending topics will not only be ranked by an algorithm, but will also include human curation. One year later, Twitter partnered We're teaming up with AP and Reuters to further our efforts to provide human-driven context to trending topics.
But shortly after Musk acquired the company in October 2022, the written context around the trending topic disappeared. Musk fired Twitter's human editor. As Reuters reported in November 2022, “Twitter's curation team, which was responsible for 'highlighting and contextualizing the best events and stories unfolding on Twitter,' has been fired.” .
Newly updated Explore page in X
Earlier this week, X-Fan accounts like @XDaily shared screenshot X's upcoming Explore page restart. His Explore page on X includes the platform's infamous trending topic list and breakouts for specific sections such as “News” and “Sports,” with trending articles by specific subject categories. to the user.
Tweet may have been deleted
Musk's updated Explore page seemed poised to bring back the written context provided to trending topics and stories.
On April 4th, X began rolling out the newly updated Explore page. This update displays top user posts on individual trending topics and an easy-to-understand summary of the subject above the user content. At the top of the page, a heading written by X provides information to readers as if they were about to read an article on that topic.
Additionally, this update adds Explore to a new prominent location on the platform. Explore's trending stories are embedded directly into the right sidebar of the main X homepage. Previously, it housed a list of trending topics viewed daily by X's hundreds of millions of users.
In a screenshot captured from popular Twitch user Hasan Piker's stream, a fake AI-generated X headline can be seen trending on the sidebar of the platform's homepage.
Credit: Hasan Piker on Twitch
But Musk hasn't brought back Twitter's curation team or hired a new human editor to write the context that X currently provides.
Context is created by Grok, X's AI chatbot.
The danger of AI-dependent X
We know that Grok makes heavy use of AI capabilities promotion by muskstrengthens X's contextual summary because X himself says so in the fine print on every Explore page.
Tweet may have been deleted
“Grok is an early feature and may make mistakes,” reads a small note from X below the context on the Explore page. “Please check the output.”
Mashable observed X's “Iran attacks Tel Aviv with large missile” page spreading fake news trending on Thursday.
According to our observations, this topic started trending when blue checkmark accounts spamming the same copy-and-paste misinformation about the Iran attack (X monthly for premium features including verification badges) This seems to be due to a sudden increase in the number of users paying subscriptions. Israel. Selected posts provided by X included a number of verified accounts spreading this fake news along with unverified videos depicting the explosion.
A side-by-side comparison of two separate posts on the X Explore page showing a fake news post from a Bluecheck user who provided information to Grok.
Credit: Screenshot from Mashable
From there, it appears that X's algorithm noticed potential story trends within these users' posts and created the Explore Stories page. From his X's own claims about the inner workings, we can infer that Grok must have created an official-looking documented story, along with a catchy headline. This was all done based on selected users sharing fake news in an attempt to automatically provide context for what the platform itself appears to assume to be the real story.
This isn't the first time Grok has provided false information to users. Previous reports on early versions of X's chatbot found that fake news was often created in his private chats with a select few users who had access to the chatbot. But the combination of this recent incident and the new His Explore feature is the first time X has taken his Grok's misinformation and packaged it as an actual trending news article, ostensibly giving users the context of the actual event. It shows that you have advertised to the entire base.
Under the Musk administration, disinformation skyrocketed With its new Explore product, the company spread falsehoods by paying bad actors to make money in return for generating engagement.
one day After Grok generated this fake story and X promoted it through its Explore page, the company made Grok available to all X users who subscribed to Premium so that they could also use the AI chatbot that created the misinformation. I made it.
Tweet may have been deleted
