The X midterm bot purge, explained: 13,000 accounts deleted

X has deleted thirteen thousand accounts it says belonged to three suspected foreign influence operations, announcing the sweep on Wednesday evening with the US midterms less than three weeks away. The company linked the networks to Iran, Russia, and China, and said the accounts used inauthentic personas to post and amplify political content aimed at candidates and public figures across the political spectrum. The X midterm bot purge is a major election-integrity move this cycle, and it arrived with a conspicuous gap: X offered no public evidence for its country-by-country attributions.

The disclosure came in a post from X's Global Government Affairs team, which framed the takedown as part of ongoing election-integrity work rather than a reaction to a single incident, according to SC Media. X said it found the accounts through proactive sweeps and investigations by its Safety team, targeting what it calls coordinated inauthentic behavior. The company gave no breakdown of how the thirteen thousand accounts split across the three networks, so it is impossible to tell whether one country's operation accounted for most of the removals. That lack of detail has followed the X midterm bot purge since the announcement.

What the X midterm bot purge left out

At the center of the story is an attribution problem. X published no examples of the content the networks amplified, no maps of how the accounts connected, and no figures on how many real users saw the posts before they were removed. WebProNews, reviewing the X midterm bot purge disclosure alongside Al Jazeera's reporting, pointed out that thirteen thousand accounts says nothing about reach, and without impression or engagement data there is no way to judge whether the operation changed any minds. The platform also named no specific candidates or figures who were targeted.

For ordinary users, what matters is what changes inside the app. X said reports filed under the civic integrity option feed into Community Notes, where contributors can add context to a post, and the company said it may apply warning labels in serious cases. The platform has not said how many people review those reports, how quickly labels appear, or whether the thirteen thousand removals came with any notice to the accounts' followers. Those operational details are exactly what researchers say they need to judge whether a purge worked, and X did not provide them.

Alongside the announcement, X pointed to its Civic Integrity policy, which covers misleading claims about how, when, and where to vote and content meant to discourage participation. Reports under the policy feed into Community Notes, the crowdsourced feature the platform leans on next to its own moderation. That reliance has drawn scrutiny: when the European Commission opened formal proceedings against X under the Digital Services Act in 2023, it questioned whether Community Notes actually works as an accuracy check. The questions followed layoffs and senior departures from X's moderation teams after Elon Musk bought the company, a detail flagged by both The Register and SC Media.

Part of a bigger pattern

The purge did not arrive in isolation. A week earlier, OpenAI said it had disrupted two influence campaigns tied to Russia and Iran that used ChatGPT to build fake journalist profiles and a phony think tank pushing anti-Ukraine messaging. The overlap in actors across the two disclosures suggests state-backed campaigns keep testing platform boundaries and move to the next service when they get caught. The X midterm bot purge fits that pattern.

The timing explains why the story is moving. Voters go to the polls on November 3 in midterms that decide all 435 House seats and about a third of the Senate, and platforms face pressure to show they can police coordinated manipulation before then. SC Media noted the accounts targeted figures across the political spectrum, which points to a goal of injecting noise into the conversation rather than boosting one side. Whether the takedown protected that conversation is harder to say. X has not explained how it attributed the networks, and until it does, the X midterm bot purge stands as both a real enforcement action and an unproven claim.