X removed 42,000 accounts that were automating replies with chatbots, and rather than announce it afterwards, the platform's head of product narrated the sweep while it ran. Nikita Bier spent about 24 hours posting live updates during the anti-spam campaign. The enforcement line he drew is the part worth reading twice: using AI to engage with users programmatically, with no human in the loop, breaks what X says its mission is.
The running commentary was written up by Search Engine Journal on 26 July 2026. Platforms almost never narrate spam enforcement in real time, and that is why the thread travelled as far as it did.
What X actually said
Bier's post on the removals: "We found 42,000 accounts automating replies using chatbots and have removed them from the platform. X's core value is providing an authentic pulse on humanity — and using AI to programmatically engage with users without a human in the loop runs counter to our mission."
He also posted: "We will clean this place. I don't care how many enemies I create. X will not be manipulated by criminals."
On timing, Bier wrote: "This spam attack was mitigated tonight. Your feed will improve in the next 6-12 hours."
The motive was money, not politics
Bier was specific about why the accounts existed. "For transparency, the bulk of them were spamming thought-leadership slop about artificial intelligence — to grow accounts and receive paid promotion offers from AI tech companies. 99.99% of spam on X is economically-motivated. Just plain old grifters."
No ideological campaign, no political operation, no state-backed influence effort. The accounts were built to accumulate an audience and then sell paid-promotion deals to AI companies. That description is worth holding onto if you buy sponsored posts from fast-growing accounts in the AI niche, because by Bier's account a meaningful share of that inventory was automated. He did not publish a breakdown of which accounts had taken money, so treat the buying implication as inference rather than something X reported.
Grok was part of the toolkit
Some of the spammers were using X's own Grok to auto-post spam replies at scale. One operator changed methods 40 times in six months, adapting each time a countermeasure landed.
The 40-changes detail says more about enforcement than the 42,000 figure does. A spam operation is a moving target with a person behind it who reads the platform's responses and rewrites the approach. Which is also why a single sweep is a snapshot rather than a resolution.
The line is the human in the loop
Nothing in the reporting says a brand account was caught in this sweep, and none was reported. But the behaviour X named, AI generating replies to users at scale with nobody reviewing them, describes a fair amount of engagement-automation tooling sold to marketing teams.
The distinction Bier drew is narrow and usable:
- Scheduling posts, drafting copy with AI, summarising mentions, triaging an inbox: a person still approves what goes out.
- A tool that watches for keywords and fires generated replies at strangers: nobody does.
Only the second matches what X described removing. If your team runs organic community management next to paid social, that is the audit question to put to every connected app: which of these can publish a message without a human approving that specific message?
What this means for marketers in Thailand
X is not the main channel for most Thai brands, and this was one platform's enforcement action rather than an industry rule. The transferable part is the stance. The rule Bier articulated targets AI-generated engagement, not AI-generated content. He did not object to AI writing anything; he objected to AI replying to people with no one in the loop.
Concrete steps that cost very little:
- List every tool with posting or reply permission on your brand accounts, and check what each one can send unattended. Revoke the ones nobody remembers connecting.
- Keep AI on the drafting side and a person on the send button. That is the configuration Bier's rule leaves untouched.
- If a vendor or freelancer reports reply volume as a KPI, ask how those replies are produced before you celebrate the number.
The same logic applies upstream in content marketing: AI-assisted drafting inside a reviewed process is a different activity from unattended reply generation, and being able to show the difference is becoming part of running a brand account safely.
Questions marketers are asking
Did X ban AI tools outright?
No. The rule Bier stated concerns using AI to engage with users programmatically without a human in the loop. His posts did not describe a ban on AI-assisted writing.
How many accounts were removed, and how fast will feeds change?
42,000, per Bier's post. He said the attack was mitigated that night and that feeds would improve in the next 6-12 hours.
Could a legitimate brand account get swept up in something like this?
No brand account was reported as affected here. The risk to watch is behavioural: automation that replies to users at scale without review looks like the pattern X removed, regardless of who owns the account.
The practical read for a brand is that reply automation now carries platform risk on X, and the cheap fix is knowing which of your tools can talk to strangers without asking anyone first. If you want a second pair of eyes on how your channels are wired, the social advertising team at Relevant Audience can review the setup.
