Introduction

“A lie can travel halfway around the world while the truth is putting on its shoes”. This line is credited to Mark Twain, but the thought is older. Jonathan Swift complained three centuries ago that falsehood flies while truth comes limping after. What has changed since Swift is not human credulity but infrastructure. The most comprehensive study of online diffusion yet conducted found that false news travels significantly farther, faster, deeper, and more broadly than the truth, and it does so through ordinary human sharing rather than automated accounts. The question here is as to what extent the state owes its citizens a duty against misinformation; cannot be resolved at the level of content alone. This post will argue that the state holds a duty that is real but narrow. It is a positive obligation to protect the conditions of safe and autonomous citizenship by regulating the architecture of virality, in the confined circumstances where amplified falsehood foreseeably threatens serious, rights-harming consequences, above all, violence.

Why a general Anti-Misinformation duty is Dangerous?

Article 10 of the European Convention is the point of departure, and it tells against a general truth-policing power. Freedom of expression protects not only the innocuous but ideas that ‘offend, shock, or disturb’, because pluralism, tolerance, and broadmindedness are constitutive of the democratic society the Convention exists to defend. Any restriction must clear tests of legality, legitimate aim, and necessity, and the same three-part discipline governs Article 19 of the ICCPR. A duty to suppress falsehood as such would strain against that architecture at its very core.

The trouble begins with definition. ‘Misinformation’ is not a stable category. The now-standard taxonomy separates misinformation (false content shared without intent to harm), disinformation (falsehoods deployed to harm), and malinformation (genuine content weaponized); yet in the moment of utterance the premature, the mistaken, the satirical, the contested and the merely heterodox are readily mislabeled. Yesterday’s proscribed claims is sometimes tomorrow’s legitimate, if unwelcome, hypothesis. A state empowered to certify truth thereby acquires an instrument that is, by its nature, available to silence dissent.

The danger isn’t speculative. The UN Special Rapporteur, surveying how states have answered disinformation, found their responses frequently disproportionate and vague, internet shutdowns, sweeping criminal ‘fake news’ offences, and themselves corrosive of the rights they invoked; she urged that any response be founded in the human-rights framework rather than substitute official judgement for public reason.

The bitter irony is that the communities most endangered by amplified falsehood are often the very ones most endangered by the over-broad laws enacted to suppress it. A general anti-misinformation duty is thus no modest reform but a standing invitation to censorship.

When should the duty arise?

If the state may not police truth at large, when must it nonetheless act? The answer is to recast the question as one of positive obligation triggered by foreseeable harm rather than by falsity. Strasbourg already accepts that the state may be required to take reasonable operational steps to protect individuals from the criminal acts of private parties. Under the Osman principle, that duty arises where the authorities knew or ought to have known of a “real and immediate risk” to life and failed to take measures within their powers which, judged reasonably, might have avoided that risk. It is an obligation of means, not of result, and mustn’t be read to impose an impossible or disproportionate burden. The analogy to misinformation is imperfect, a platform is not the state, and a falsehood is not a knife, but it is illuminating, for it fixes the trigger in concrete, foreseeable danger rather than in the wrongness of an idea.

On that foundation, the state’s duty should crystallize only where four conditions converge. First, the content must be demonstrably false or seriously misleading, not merely contested, unpopular or premature. Secondly, it must be amplified at scale: it is reach, not error, that turns a lie into a public danger, so the wrong to be addressed is systemic rather than the isolated utterance. Thirdly, the amplified falsehood must be tied to a foreseeable risk of serious rights harm, to life and bodily integrity, to equality and freedom from discrimination, to public order, or to democratic participation. Fourthly, any intervention must be capable of proportionality: sufficiently targeted and time-limited to meet the danger without smothering lawful debate. What distinguishes the cases that should engage the state isn’t that a falsehood exists but that platform design, recommender ranking, monetized virality, frictionless resharing, has manufactured its reach.

International practice already furnishes the analytic tools for the crucial third limb. The Rabat plan of Action, devised to discipline incitement law under Article 20(2) ICCPR, sets a deliberately high-six-part threshold, context, speaker, intent, content and form, extent of dissemination, and likelihood of harm including imminence, before expression may lawfully be curtailed. Two of those factors, the extent of dissemination and the likelihood of imminent harm, map almost exactly onto the amplification and foreseeability core proposed here.

The summer of 2024 displays the test in motion. Within hours of the murder of three children at a dance workshop in Southport, a fabricated name and the false assertion that the killer was a Muslim asylum-seeker who had crossed the Channel by small boat raced across social platforms. The false name attracted tens of thousands of mentions before the suspect’s name could lawfully be published; Merseyside Police had already confirmed that the suspect was born in Cardiff, and later confirmed that the circulated name was false. The claim was demonstrably false, amplified at scale, and tightly bound to a foreseeable risk of racialised violence. That risk duly materialised as mosques, asylum accommodation and migrant communities were targeted across England. That is precisely the narrow zone in which a state duty arises. Equally, the test would not be satisfied by a contested scientific claim or an unpopular political opinion circulating without that conjunction of scale and foreseeable violence, which is the point of drawing the line there.

Gap in the UK’s Law

Measured against that standard, the UK’s law is incomplete in a telling way. The Online Safety Act 2023 stated that it is a crime to send a message conveying information the sender knows to be false, intending thereby to cause non-trivial psychological or physical harm to a likely audience, and without reasonable excuse. At the individual, criminal margin the offence is genuinely useful: some post-Southport unrest communications were charged or prosecuted under it, including a TikTok livestream falsely claiming pursuit by rioters. But its reach is narrow by design. It demands knowledge of falsity and an intention to harm, so the merely reckless re-sharer who gives a lie its decisive velocity will usually fall outside it. Recognised news publishers are exempt. And although the new offence has a wider drafting reach in some respects, it largely replaces the older false-message offences it superseded. None of this directly touches amplification.

The structural gap is lawful-but-harmful misinformation propelled by recommender systems. Parliament’s Science, Innovation and Technology Committee, reporting on 11 July 2025 after an inquiry prompted by the riots, judged the online-safety regime to contain “major holes” and to be “already out of date”. Its central finding was architectural: platform business models incentivise the spread of damaging and dangerous content and did so in a manner that endangered public safety in the days after the Southport murders. The Committee also found that the Act was not designed to tackle misinformation and would have made little difference to the misleading content that drove violence and hate in summer 2024. The few measures in the Act that address misinformation therefore fall short where the content is harmful yet not illegal. The Committee accordingly recommended duties to risk-assess and report on legal-but-harmful content with a focus on recommendation algorithms, and to identify and algorithmically deprioritise fact-checked misleading content where it has the potential to cause significant harm: regulation, in substance, aimed at virality. That the original adult “legal but harmful” duties were stripped from the Bill amid free-expression concerns means the gap was not simply an oversight but a deliberate, and now costly, choice. The point remains even after Ofcom’s 2026 crisis-response measures: those measures strengthen protocols for spikes in illegal content and content harmful to children, but they do not create a general duty to decelerate lawful harmful misinformation addressed to adults.

The conventional alternative, media literacy, is valuable but mistimed. Teaching citizens to interrogate sources is a sound long-term civic investment. Indeed, the Act specifically expanded Ofcom’s media-literacy duties to include awareness of misinformation, disinformation, reliability, accuracy and authenticity. But media literacy cannot operate at the tempo of an acute crisis, when a false name is already trending, and a crowd is already assembling. The Committee itself found that the media-literacy strategy did not set ambitious goals. More fundamentally, psychological research on misinformation shows that corrections and critical-literacy interventions face persistence, repetition, identity and fluency effects; they are necessary civic infrastructure, not emergency brakes. A duty that bites only over years cannot answer a harm measured in hours.

A Better Model: The Architecture of Virality

The remedy is not deletion-first content control but what might be called an epistemic fire brigade: fast, temporary, transparent and appealable interventions trained on amplification rather than on truth. The European Union’s Digital Services Act supplies the template. It obliges the largest platforms to assess the systemic risks arising from the design, functioning and use of their services, expressly including negative effects on fundamental rights, civic discourse, electoral processes and public security. It then requires reasonable, proportionate and effective mitigation measures tailored to those risks. It couples that risk-management model with a crisis-response mechanism, crisis protocols, independent audits, transparency reporting and mandated researcher access to platform data, so that intervention is both timely and auditable.

What renders the model compatible with Article 10 is the red line it observes. As Husovec demonstrates, the DSA does not empower the regulator to invent new content rules or to dictate which lawful expressions a platform must suppress; it governs systems and risk, not truth, which is exactly why the feared Ministry of Truth fails to materialise. The duty lies on the architecture of virality, not on the citizen’s right to be wrong.

Rendered into a domestic duty, that implies a defined and narrow toolkit: crisis protocols capable of activation during acute, violence-adjacent events; transparent and reversible downranking, not removal,  of fact-checked false content meeting the four-part test; friction introduced before resharing at the moment a cascade accelerates. Prominent public-interest corrections and independent fact-checking should be deployed to meet speed with verified counter-speech. Researcher access, algorithmic audit duties and regular reporting to Parliament should make the system externally testable rather than privately self-certified.

Every measure is procedural and structural. None asks the state to declare what is true; each slows the machinery by which falsehood becomes foreseeable violence. To answer the objection of Part I, the trigger and its application must rest with an independent regulator and be subject to prompt, effective appeal, never with the executive of the day; it is transparency, reversibility and external oversight that keep an emergency measure from hardening into a censor’s charter. The object is to add friction to amplification, not to adjudicate ideas, and friction, unlike prohibition, leaves the marketplace of ideas open while denying falsehood its artificial, engineered velocity.

Conclusion

The honest answer here is one of degree and of direction. The state owes its citizens no general duty to shield them from error, and a state that crowned itself the arbiter of truth would betray the very expression it professed to protect. But the state does owe citizens a duty not to stand idle while private architectures of amplification convert demonstrable falsehood into foreseeable, rights-harming violence. That duty is structural, not doctrinal; it is directed at virality, not veracity. Not a Ministry of Truth, but not a bystander either.

Aaditya Bajpai is a Master of Law Student at the University of Edinburgh.

(Suggested citation: A Bajpai, ‘Not a Ministry of Truth: The State’s Positive Duty to Prevent Algorithmically Amplified Misinformation from Triggering Rights-Harming Violence’, U.K. Const. L. Blog (2 July 2026) (available at https://ukconstitutionallaw.org/))