15.01.2026

No trust in the state

Despite Elon Musk’s climbdown, there is still talk of further legislation to restrict X and other social media platforms. But should the left trust Keir Starmer and Liz Kendall when it comes to policing online content? Paul Demarty thinks not

It is strange, but all too predictable, that - of all the possible things to pick a fight with the United States over - our government has chosen AI-generated porn.

There has been a spate of cases where users of X - Elon Musk’s rebranded Twitter - have used the app’s built-in AI chatbot, Grok, to create lewd images. Grok can be instructed to remake any image on command. It is impossible to simply instruct it to make a woman in a picture naked, but, as always, carefully crafted prompts can work around such guardrails. You can tell it to put the woman in a bikini. Then, someone discovered, you can ask it, in various ways, to render the bikini extremely thin, or even demand a bikini made out of dental floss, and so forth.

In some cases, no doubt, this is used by some of the world’s saddest men for masturbation. In others, it is simply another weapon in the endless war of all against all that is online political argument. Women caught in the act of posting progressive opinions can expect to find themselves dressed up in itsy-bitsy, teeny-weeny bikinis by cretinous rightwing trolls. Sometimes their bodies are bruised, or they are bound and gagged. It is, to be sure, not a terrifically edifying spectacle.

Orwellian

Since this activity violates a host of laws in the UK, including the recent Orwellian Online Safety Act, there is a push to take action against the app that many still call the ‘hellsite’. Ultra-Blairite technology minister Liz Kendall has prodded Ofcom into investigating, and ministers openly threaten a ban. For now, Musk strikes a defiant pose, accusing Starmer’s government of “fascism”, which - given the general political atmosphere on X these days - is a bit rich.

It should come as no surprise to regular readers of this paper that we oppose further restrictions on online speech, just as we have always opposed clampdowns on online harassment, in the name of preserving liberty of political expression at large. Because the particular phenomenon at issue in the present context is grossly degenerate and morally contemptible, that is not a compelling reason to break from this policy: after all, there is always some pressing reason for action, and it is rarely completely made up. That said, it does raise particular questions: what, if anything, is to be done about this and other issues of ‘online safety’? And - to be blunt - why are people on the internet such arseholes?!

To take the ban first, there is a simple general argument against restricting free expression in the name of safety, even in the name of the safety of those who are oppressed in some way or another. It can be posed most clearly with respect to criminal sanctions against, say, hate speech. When such a law is on the books, the problem is posed of determining whether some particular utterance or online post is ‘hateful’ to the appropriate criminal standard. This duty will always fall to the police and prosecutors. The question of whether hate speech should be criminalised, therefore, is inescapably reduced to the question of trust in the repressive apparatuses of the state.

Now, we are not precisely in the same situation with the X ban. Here there are not police involved in the literal sense, but rather the regulatory body, Ofcom. Yet the underlying issue is the same: Ofcom, after all, is no more under the control of the general population than Scotland Yard. Its decisions are made in the interests of the state.

It is the job of Marxists, however, to cultivate distrust of the bourgeois state, to erode - so far as possible - its legitimacy as an arbiter between contending forces in society, and ultimately replace its bureaucratic and repressive institutions with democratic means of administration and justice. If we trust the state to determine the acceptable limits of discourse, then we invite such perverse results as the near-criminalisation of pro-Palestinian agitation as ‘anti-Semitic hate speech’. Ofcom, with its broad purview over media content, also has plenty of room to make mischief.

The ban has its more absurd aspects, too. After all, it would not actually prevent the general run of X users - overwhelmingly outside the UK - from making these images. It would merely prevent British victims from seeing them, and even then only British victims who lack the minimal technical competence required to sign up for a virtual private network (VPN) service. It makes sense, instead, only as part of a general drift towards regimes of generalised censorship, summed up in laws like the Online Safety Act.

The political right has set itself up as the ‘resistance’ to such censorship in recent years. Indeed, the Trump administration itself has repeatedly warned European powers against attacking ‘freedom of speech’. The matter even showed up in the administration’s recent national security strategy document. Liberals and the left rightly point out that this is all laughably hypocritical, and the second Trump administration has so far proven itself far more censorious than any recent Democratic regime.

Yet the insincerity is not a reason for the left to dismiss concerns about censorship; if anything, it is the opposite. Starmer and co are assembling repressive institutions that, at the next time of asking, look nailed-on to be handed over to Nigel Farage or a suitably ‘Faragified’ Tory leader. That, alone, would be reason for concern. (Of course, Starmer is quite repressive and cop-brained enough himself already …)

Abuse

If broad bans are to be opposed, where does that leave us with respect to these images? It should be said, first of all, that the bikini pics exist on a continuum with other kinds of online ‘abuse’ - by ‘abuse’ I mean merely interventions in public discussion that attempt to obtain victory other than by means of substantive persuasion. Instead, the ‘abuser’ attempts to construct, with apologies to Theresa May, a hostile environment for their opponent, such that the latter will simply drop out for the sake of their sanity. In this respect, there are many tactics available - mob-handed, mass denunciation (popular on the liberal left in the 2010s); death and rape threats (more popular on the right); doxxing (fairly universal); and so on.

For some such tactics, a criminal sanction is all but inescapable. A good example would be ‘swatting’ - the attacker makes a hoax report of a hostage situation at the target’s home, hoping to trigger a police raid. A similar case could be made for actual revenge porn (the distribution of graphic images of the target), especially where that material has been obtained by unauthorised access to their private accounts. Serious death threats - where the attacker plausibly could carry out the threat - likewise.

Yet much else besides is simply not surgically distinguishable from fair comment. Take the common practice of exhorting one’s opponent to commit suicide. Can this be rigorously distinguished from assertions of the form, ‘If this politician lived in feudal Japan, he would have to commit seppuku’ - or, indeed, the perfectly defensible statement (though I disagree with it as an opponent of the death penalty, even in such cases) that the architects of the Gaza genocide, or some other act of mass murder, should be tried and hanged? Though it is undoubtedly psychologically harmful to maintain an ever-growing kill list in one’s mind, there cannot, in other words, be a right not to be wished dead without drastic infringements on discussions on - in this case - the legitimacy of the death penalty and the demands of personal honour. Remember that all this will be decided by PC Plod.

Synthetic revenge porn is more similar to the mob-handed denunciations of old. Both are effectively forms of vigilante justice - attempts to expel someone who has transgressed some norm by way of terrorising them. The porn images are, additionally, characteristically misogynistic (though men are sometimes the target: far-right trolls succeeded in inducing Grok to write an elaborate and extremely violent rape fantasy about the American liberal pundit, Will Stancil), but the social function of the act is similar.

It cannot be argued by any supporter of freedom of association that it is illegitimate for social groups to police their boundaries, and in any case it would be fruitless: such boundaries are probably ineliminable from human nature, never mind organised political disagreement. So who is to decide what is a legitimate or illegitimate method of doing so (excepting direct violence or other unproblematically criminal activity)? Again: PC Plod.

Bad culture

In fact, we need to take a wider view to get a grip on the problem, and we have to start by acknowledging that abuse is not really aberrant on modern social media: instead, it is the norm. The standard of political argument on platforms like X is abysmal. I have already had cause to refer to “the common practice of exhorting one’s opponent to commit suicide” and, when seriously opposed viewpoints are contested, that is about as good as it gets, most of the time. Some worthwhile discussion happens among political near neighbours, but even that can readily degenerate into exchanges of barbed one-liners or mass denunciations, most absurdly in the ‘circular firing squad’ scenario (A denounces B, B denounces C, C denounces A).

The more radical question is thus: why? There is an implicit answer that comes with the speech-policing approach to particular cases, and it is a fundamentally conservative answer. It takes for granted that this is, in some respect, simply a feature of how we are. As John Calvin said, the human condition is one of total depravity. Only the threat of sanction can get us to behave.

If this conservative view is true, then the socialist project is doomed, and we merely await the ultimate circular firing squad: generalised nuclear exchange. Of course, we can hardly deny that some people have it in them to behave in these ways: the evidence is before our eyes. Yet we need an alternative, and better, interpretation of these facts.

We can start from the commonplaces of modern media criticism. Social media is governed by the selections of the algorithms, and the algorithms are determined by the needs of the platforms as capitalist enterprises. They make money from showing adverts, and therefore their users must spend as much time as possible scrolling their feeds and looking at those adverts. Anxiety and rage does the trick better than sunshine and puppies, and so these platforms are characteristically unhappy places.

Perhaps more than the algorithms, though, there is the overall structure - millions of individual users in a single discursive space, their interactions mediated primarily by the platform itself. This is a recipe for atomisation, and consequently forms of political culture characterised by personal loyalty to favoured celebrities. The greatest example is, of course, Donald Trump, but very much smaller fry can become the object of such investment for smaller groups (vide Kamala Harris and her ‘KHive’ superfans).

Escaping this degenerate culture, then, is a matter of constructing alternative institutional forms that cut against atomisation and its necessarily Bonapartist consequences. Political comradeship and opposition would then not be mediated by the particular corporate interests of Elon Musk or Mark Zuckerberg, but by organisations built for the purpose - democratically organised parties, in short.

Yet the capitalist class has no need of parties in this sense. Indeed, mass-membership bourgeois parties largely exist in response to the parties of the one class that has only coherence in large numbers to rely on: the working class. It is our job, in other words, to provide the alternative, both as a programme for social transformation and as a living body that, somehow, models a better and democratic culture of political contestation.

We cannot leave it to Ofcom, or Musk, or PC Plod!