Nikita Bier, head of product at X (formerly known as Twitter) sent a series of unusual tweets about spam on Twitter, explaining the motivation of some spammers, explaining the types of spam and saying that some spammers use Grok to automatically send spam responses at scale.
Nikita Bier spent 24 hours tweeting anti-spam actions in real time, detailing how automated and fast AI-assisted platform spam has become.
42,000 Accounts in One Sweep
Bier began their series of tweets explaining the scope of the chatbot problem, saying that providing authenticity to X is a core value at the core of the company’s values.
Bier tweeted:
“We found 42,000 accounts’ automated responses using chatbots and removed them from the platform.
X’s core value is providing a true human touch – and using AI to systematically engage with users without a human in the loop is the opposite of our mission. “
Chatbot Spam is Monetized
Bier followed up the next day with an explanation about what chatbot spam was all about. They explained that the spam was not ideologically driven, it was not a political or government operation. It was only about making money. Apparently the spammers were trying to grow the audience in order to leverage that into paid advertising deals from AI companies looking to expand their influence.
Companies rarely post about their spam problems and Bier’s real-time postings provide a behind-the-scenes look at the scale and motivations.
Bier tweeted:
Maybe you should get a different app pic.twitter.com/z9Qv0vbpZj
— Nikita Bier (@nikitabier) July 24, 2026
And he posted an explanation:
“To be clear, a lot of it was thought leadership spam about artificial intelligence – to grow accounts and get paid promotional offers from AI technology companies.
99.99% of spam on X is economically motivated.
Just plain old grifters. “
Status Update: A Fix in Hours, Not Days
A day later Bier posted a quick update on the timeline of when you might see X feeds develop, and it shows how these types of anti-spam actions work in real time.
Bier tweeted:
“This spam attack has been reduced tonight. Your feed will improve in the next 6-12 hours.”
UX Fights Back Against AI Chatbot Spammers
Bier’s tweets served as an open statement to spammers to show how serious they are about fighting fake behavior on the platform, calling spammers criminals.
Bier tweeted:
“We will clean this place, I don’t care how many enemies I make, X will not be used by criminals.”
Spam Voted 40 Times
The most eye-opening part of the series wasn’t a brief exchange between Bier and one of the X members. Bier described one spam operator who was so persistent and flexible (40 method changes in six months), that the person sending the spam behaves less like a bot farm and more like a person sitting next to him who keeps track of all his responses and quickly comes up with countermeasures.
A response from member X drew attention to the fact that some of these spammers have started using Grok, X’s AI product, to generate their responses. Bier confirmed that the approach based on Grok has already been intercepted and blocked. In their response, Bier pointed out that X’s 12-18 hour turnaround time was an improvement over how long a similar problem was taking under Twitter’s past.
Bier tweeted:
“There are quite a few spammers on X who have been using their strategy for the past 6 months.
One of them (“This guy is a great trader ⬇️”) pointed 40 times after each path was blocked.
Some techniques are so clever and fast that it feels like they are sitting right next to us.
In the meantime, we might hire them because they are as familiar with the X codebase as we are.”
@CryptoParadyme responded:
“Please don’t encourage them
many of them have been using some form of grok feedback lately. “
Bier tweeted:
“We blocked Grok’s yesterday.
Our team is waiting for their next move. We are 10 times more efficient than before.
This will increase for months on Twitter but our turnaround time is now 12-18 hours. “
False positives were reported and dealt with
Another interesting result of this thread about spamming in real time is that one person posted about his negative experience with spamming. A false positive is when the machine makes a mistake, labeling something as spam when it actually isn’t. In real life, these types of algorithms rely on a large number of signals to identify spam with a high degree of accuracy but false positives still occur, almost a low percentage of the time.
@the_defi_dad tweeted:
“I have to be honest. I was surprised to get a notification that my account was spam. I clicked the review request button and after 12 hours I was called back. I complained that the grass app was a scam and I got a notification right away. I’m not sure of the link there but I’m glad to see the algorithm or whatever that decides spam or no spam made the right call.”
Bier’s post received a positive response from X users, some of whom expressed that some of these discussion forums should generate engagement numbers but that their authenticity makes them a disaster.
@RandomPerson242 tweeted:
“On the one hand, people kind of love this garbage. On the other hand, I agree that you can’t let that be the content of the site, even if people follow it. It can’t be a race to the bottom.”
Social Media Is Best When It’s Authentic
Many were happy to see that X was fighting back against real AI chat spam. It destroys the experience because people come to social media like X to learn and share people’s experiences.
Featured image by Shutterstock/Naumova Marina
