The Great Paradox of X: Politicians Want to Ban Musk… But Can They Still Do Without Him?
It was a hot summer—on X as well. We witnessed some interesting exchanges centered on the question “Should X be banned?” For Axel Legay, this question masks a far more awkward one: “Why do those who want to ban X continue to need X?”
I leave, I come back, I leave, I… More prosaically, can we fight a power infrastructure while still needing that infrastructure to exist politically?” asks Axel Legay, an expert in cybersecurity, artificial intelligence, and digital governance.
This summer, we saw several prominent figures take a stand. Bernard Arnault on July 27, Jensen Huang on the 24th, Mark Zuckerberg on the 9th: in just a few weeks, several executives broke their silence or made their debuts on X. It remains to be seen whether this trend reflects a genuine return of the platform to the economic debate or a series of isolated cases.
Three executives, three different contexts: a personal response for Bernard Arnault, an industry-wide stance for Jensen Huang, and a product launch for Mark Zuckerberg. Does the choice of X as a platform for expression signal a shift in how business leaders engage with the public sphere?
Why is X so controversial?
There was a time when Twitter almost resembled a digital public square. Twenty years later, the platform’s philosophy has changed. Elon Musk, its current owner, has made the defense of free speech one of the central arguments for his takeover of the platform. His critics, on the other hand, see it as a rollback of the mechanisms that helped limit hateful, misleading, or manipulative content.
The first criticism concerns moderation. But reducing the debate to a dichotomy between “moderation” and “freedom of expression” would be too simplistic, according to Axel Legay. “The real issue is the governance of an information space frequented simultaneously by citizens, the media, politicians, businesses, activists, and organizations that sometimes deliberately seek to manipulate the debate. ”
Who decides what can be published? Who decides what will be seen most of all? This second question is probably even more important than the first.
Why some content is visible… and other content less so
X has made the openness of its recommendation algorithm a selling point for transparency. Part of the code has indeed been published and documented. In August 2026, X further strengthened this approach by publishing the code for its Phoenix recommendation system.
“But ‘publishing the algorithm’ doesn’t necessarily mean allowing an outside observer to exactly replicate what a user sees on their screen,” Axel Legay clarifies. A modern recommendation system isn’t limited to just a few lines of code. It depends on the models actually deployed, the data used to train them, production settings, behavioral signals, ongoing experiments, and changes introduced between versions.”
In other words, one can understand part of the mechanics without being able to precisely observe the machine in operation. Moreover, this problem is not merely theoretical. In January 2025, the European Commission asked X to provide internal documents regarding its recommendation systems and their recent changes as part of its investigation under the Digital Services Act. In December of the same year, it fined X 120 million euros for several violations of the DSA’s transparency obligations, particularly regarding data accessible to researchers and the advertising registry.
For Axel Legay, the question is therefore no longer simply, “Is the algorithm public?” It has become, “Are we truly capable of understanding why some content becomes extremely visible while other content remains practically invisible?”
When the network’s owner becomes a political actor in his own right
With X, another unique aspect makes this issue particularly sensitive: its owner has himself become a major political actor.
Elon Musk doesn’t just manage the platform. He intervenes directly, has a massive audience, and regularly takes a stand on elections, governments, political parties, or political figures. “This situation creates an unusual dynamic. The owner of the infrastructure is simultaneously one of its most influential users. ”
Moreover, the question is no longer simply whether Musk personally benefits from X. It concerns the recommendation system’s ability to alter users’ exposure to certain political discourse.
A study published in February 2026 in Nature, based on an experiment conducted on X in 2023, offers a particularly interesting insight in this regard: the researchers conclude that using X’s algorithmic feed can indeed alter exposure to political content and produce measurable effects on certain political attitudes.
“This does not allow us to conclude that every political outcome observed on X is deliberately programmed by its owner. That would be going much too far. However,” continues Axel Legay, “it does show that the algorithm is not merely a technical detail.” ”
The architecture of a social network is also an architecture of visibility
And when a platform owner becomes directly involved in political debate, transparency regarding this architecture becomes a democratic issue.
Recent events also demonstrate just how irrelevant national borders are becoming on these platforms. For example, an analysis published in August 2026 on discussions surrounding the unrest in Belfast and Southampton found that fewer than half of the accounts that generated more than 1,000 views on these events were located in the United Kingdom.
“A local controversy can now be amplified by thousands of actors who have no connection to the area in question,” concludes Axel Legay. X is therefore no longer simply a network that allows politicians to communicate. X has become one of the arenas where the relationship between politics, information, and influence is shaped. ”
Why Do Politicians Stay on X?
In the face of these controversies, one question comes up regularly: why not just leave X? Some have done so. Others have scaled back their activity. Alternatives like Bluesky, Mastodon, or Threads have attracted new audiences. But leaving X individually is much easier than replacing X collectively.
The network’s main advantage isn’t technological. It’s its networking effect. “A minister stays on X because journalists are there. Journalists stay because politicians, experts, and companies are there. Companies post there because their customers, partners, and the media might be there. And everyone keeps their account because they’ve sometimes built up tens or hundreds of thousands of followers over ten or fifteen years.”
These relationships constitute digital capital that is difficult to transfer. “Creating an account on another platform takes a few seconds; rebuilding ten years’ worth of following there can take years,” points out Axel Legay. “That’s probably the best protection X has today!”
Can X really be replaced?
Technically, yes. Socially, it’s less clear-cut. There are now several platforms capable of replicating a large portion of Twitter’s historical features. Some even offer more open architectures or different approaches to moderation. But no migration automatically transfers journalists, politicians, institutions, experts, and—above all—their millions of connections with their followers.
Perhaps that is the paradox of X, Axel Legay observes. “Never has the platform been so hotly contested. Its moderation, its recommendation system, its transparency, and its owner’s political involvement are now the subject of a debate that goes far beyond technology. And yet, those who criticize it often continue to use it… Replacing the technology may be relatively easy; replacing a network, much less so.”
