It appears that X, the company formerly known as Twitter, has a problem with the Verified bot. Although X owner Elon Musk suggested that forcing users to pay for verification would help eliminate bots (aka automated accounts) on the platform, that doesn’t seem to be the case. ONE video gaining views on the rival platform Instagram Threads shows X search results where many bots, including many verified with a blue check, post a variation of the phrase “Sorry, I can’t answer as it is against the OpenAI use case policy. “
The response is what OpenAI’s chatbot says when a user asks a question or requests to perform a task in violation of OpenAI’s terms of service. In this case, it’s also an indication that the X account in question uses artificial intelligence to generate its posts.
A video of these search results was posted on Threads by a Chicago-based writer and former Media Matters editor Parker MolloyWHERE he noticed“Twitter is a ghost town.”
Posted by @parkermolloy
View in threads
Musk, by the way, is suing the nonprofit Media Matters alleging defamation after the publication article which showed screenshots of ads appearing alongside hate speech on the platform. The suit comes amid an advertiser exodus that is undermining X’s potential revenue, prompting Musk to quip that departing advertisers should “fly” themselves, in an interview about the issue at The New York Times’ DealBook Summit at the end of November.
If you can’t see Molloy’s video on Threads, you can see the X search results firsthand by searching for the AI chatbot’s answer in quotes, as in this question here.
In the comments posted in response to the video, a user behaves that the bot’s activity could be coming from X itself. They argued that many of these blue-tagged accounts appear to be older, abandoned accounts that were reclaimed and then turned into blue-tagged audits that use artificial intelligence to automate the account. This could be an effort to boost metrics like daily and monthly active users, they suggested.
It appears that at least some of the bot accounts are older, according to the “join date” shown on their X profile. You can see one example from this here, for example (see below). These accounts also post content that reads like it’s the output of some AI query, which it probably is.
Of course, this doesn’t prove that the account is run by X himself, rather than a spammer squatting on an abandoned Twitter handle.
Still, it’s suspicious to say the least, and it certainly proves that paid verification alone isn’t a solution to X’s bot problem — an issue Musk solved when he acquired the social network more than a year ago. More recently, he suggested that everyone should pay for X, as a “small monthly payment” would help fight the “vast armies of bots” on the platform.
Despite the numerous posts from these bots, AI-powered accounts are not X’s only problem. Many bots and bot farms operate without the help of OpenAI and are harder to detect. According to data obtained from Fedicaa social media analytics and publishing platform, only 202 accounts published OpenAI’s automated response in the last 30 days; as seen in this question here. While some were from real people joking about the bot problem, the majority were AI responses. More bots may have already been deleted from X, but this data is not available.
It appears that the Verified bots are largely X accounts created in the November 24-26, 2023 time frame.
This isn’t the only area where X struggles with bots. The company admitted last summer that it had a Verified spammer problem when it announced the new DM settings. The new features were aimed at reducing spam in users’ IM inboxes by moving messages from verified users you don’t follow back to the Message Request inbox — another sign that X’s verification system isn’t weeding out the spam as hoped.
X did not respond to requests for comment on this matter.
The company claims to have 550 million monthly users, per Musk, and the company sees 500 million posts a day, including posts, replies, posts and reposts, according to X CEO Linda Yaccarino. However, neither performer has said whether the bots took those metrics into account.