Big Brother Watch, Britain’s Free Speech Problem: How Growing Restrictions Are Stifling Speech Online and Offline, London: Big Brother Watch, published 7 September 2026, 104pp.

Big Brother Watch has published a report that everyone concerned with the condition of liberty in Britain should read. This includes people on the Left who discovered civil liberties when the police began arresting supporters of Palestine, people on the Right who discovered them when immigration sceptics were visited at home, and those of us who had noticed rather earlier that giving the State vague powers over speech was unlikely to end well.

Britain’s Free Speech Problem is a long, carefully documented account of how successive governments have narrowed the space in which British people may speak, protest and associate without official interference. Its subjects include the Online Safety Act, the criminal law governing electronic communications, the restriction of demonstrations, and the secretive machinery by which Whitehall monitors what it calls disinformation. These may look like separate questions. The value of the report lies in showing that they are parts of the same development.

Britain has not acquired a Ministry of Truth. It has done something more characteristically British. It has accumulated regulators, statutory duties, police guidance, ministerial units, content classifications, protest conditions and vaguely worded offences. Each measure was presented as a reasonable answer to some immediate difficulty. Each arrived bearing an assurance that no decent and law-abiding person need worry. Their combined effect has been to establish the principle that lawful expression is no longer presumptively free. It is something to be balanced, supervised, risk-assessed and, where convenient, hidden behind an identity check.

Before reaching the excellent substance of this report, I must deal with its extraordinary foreword by Nadine Dorries. She was Secretary of State for Digital, Culture, Media and Sport while the Online Safety Bill was being driven through the House of Commons. She now appears at the front of a report documenting the resulting censorship, presenting herself as a disappointed friend of liberty. The confession is useful. Almost everything surrounding it is intolerable.

Mrs Dorries begins by declaring that she “will always defend” the supposedly “legitimate, important and honourable objective” behind the Act. It was intended, she says, to protect children from abuse, grooming, pornography, self-harm and suicide. What it was not intended to do was limit freedom of expression “as in some cases, it appears to have done” (p.4).

Appears to have done.

The report that follows documents closed forums, compulsory age checks, demands for identity papers, blocked political videos, restricted news reports and ordinary conversations hidden from British users. It records a system in which adults must sometimes submit a passport or an image of their face before reading lawful material. These are not spectral effects glimpsed through ministerial fog. They are the predictable consequences of legislation that Mrs Dorries promoted.

Her attempted explanation is worse. The Bill, she says, was “originally meticulously drafted”, but “numerous MPs from different parties” tried to “piggy back” on it with their “pet projects”. The legislation was thereby “distorted” until it failed both to protect children and to preserve freedom of speech (p.4).

How convenient. Mrs Dorries and her Conservative colleagues created the machine, rammed it through Parliament beneath a banner saying “Protect the Children”, and supplied the regulator with its coercive power. Now that the machine is behaving as its critics warned, responsibility belongs to an anonymous crowd of meddling MPs. Apparently, the Secretary of State was present for the legislation’s passage but absent from its contents.

Mrs Dorries was not a junior clerk who found the finished Act on her desk. She was the Cabinet minister responsible for the policy. She had the departmental apparatus behind her. She participated in the moral intimidation by which legitimate objections were treated as indifference to injured children. If the Bill was ruined by amendments, she had a duty to resist them. If it could not be rescued, she had a duty to withdraw it. If she lacked the authority to do either, she had the option available to every minister who discovers that his government is about to injure the country: resignation.

She did none of these things. She now tells us that the legislation became “the digital equivalent of a dog’s dinner” (p.6), as though she had merely ordered something unfortunate in a restaurant.

There is a stale convention in British public life that failure is expiated by retrospection. A minister introduces a ruinous policy, spends years denouncing its opponents, and then returns after the damage is undeniable to explain that it was never supposed to happen. If the apology is sufficiently wistful, he may even be praised for candour. The public is invited to treat the confession as a public service rather than an admission of guilt.

I do not accept this convention. The people who construct instruments of censorship should not be allowed to rehabilitate themselves by contributing forewords to reports about the censorship. Mrs Dorries’s confession has evidential value because it shows that even one of the Act’s principal political sponsors can no longer deny the outcome. Beyond that, her presence almost undermines the report’s credibility. It creates the impression that those responsible for the disaster may convert responsibility into a speaking engagement and then join the victims in deploring what happened.

I do not suggest that Mrs Dorries committed treason as that offence is presently defined in law. I say something politically more important: she and her Conservative colleagues betrayed the liberties they were entrusted to preserve. If political treachery against the British people were treated with the seriousness it deserves, they would be answering charges, not writing apologetic forewords. At the very least, no one who helped force this legislation through Parliament has earned the right to describe its censorship as an “unanticipated effect”. It was anticipated. The warnings were public, specific and repeated. The ministers chose not to listen.

There is one sentence for which Mrs Dorries deserves credit: “Children deserve protection. Adults deserve liberty” (p.6). This is correct. It would have been more useful when she was in office.

The first major section of the report explains what her Act has done. The Online Safety Act requires Internet services to prevent children from accessing specified categories of material and to control access to other content deemed harmful. Its reach extends far beyond the enormous social-media companies invoked during parliamentary debate. It applies to user-to-user services generally, including small forums maintained by volunteers and communities with neither a compliance department nor a resident barrister.

Ofcom estimated that more than 100,000 services would fall within the system. For Meta or Google, regulatory compliance is another cost of business. For a discussion forum devoted to hamsters, cycling or life in an Oxfordshire village, it may be the difference between continuing and closing. This is one of the report’s most valuable findings. Regulation advertised as a restraint on Big Tech has strengthened Big Tech by imposing costs that its smaller competitors cannot bear.

The case of Microcosm makes the point. The platform supported about 300 small forums. Its operator, Dee Kitchen, obtained legal advice and concluded that the risk assessments, technical alterations and possible personal liability created by the Act were too onerous. The servers were closed in March 2025. Only two or three of the communities appear to have survived in anything like their original form. Some groups moved beneath the shelter of larger platforms such as Discord. An Act supposedly designed to discipline the technology giants therefore drove independent communities into their arms.

This is not simply a collection of unfortunate anecdotes. As the report observes, “Every forum that closes reduces the diversity of online discussion. Every community that limits the subjects its users can discuss narrows the space for free expression” (p.22). That is exactly right. A free Internet is not defined merely by the continued existence of half a dozen multinational platforms. It depends on the ability of small groups to establish and maintain their own spaces without first constructing a compliance bureaucracy.

The same fate overtook Dads With Kids, a forum for single and separated fathers. It had more than 2,600 members and contained about 27,000 posts in 3,700 discussion threads. Its administrators closed it because the new requirements were too demanding. A small remnant migrated to Discord, but most of the community disappeared.

There is a human reality behind this. A father sitting alone after a divorce may not want a national organisation, an approved charity or a therapeutic intervention. He may want to ask other fathers how they arrange Christmas, respond to a frightened child or manage a dispute over contact. The State did not prohibit these discussions. It merely made maintaining the place where they happened sufficiently troublesome that the place closed. Modern censorship often works in this fashion. The door is not nailed shut by a policeman. The owner receives a compliance manual and closes it himself.

This also reveals the class character of regulation. Large companies can employ teams of lawyers, moderators and public-policy specialists. Small associations depend on spare time and goodwill. A uniform legal duty is therefore nothing of the kind in practice. It bears lightly on the corporation and crushes the amateur. The village noticeboard, the specialist forum and the self-help group are treated as miniature versions of Facebook, then blamed for lacking Facebook’s resources.

The report is equally good on age verification. Since the Act’s child-safety provisions came into force in July 2025, British users have encountered demands for proof of age before accessing lawful material. This may involve an identity document or a biometric estimate derived from the user’s face. Those unwilling to submit are offered a restricted Internet.

The familiar answer is that nobody is compelled to use a particular website. This evades the point. When the law creates the commercial incentive for platforms to demand identification, the resulting loss of anonymity is a consequence of State policy. Moreover, an estimated five million British adults lack a standard form of photographic identification. A legal structure that conditions access to information on the possession and disclosure of official identity papers has created something more serious than an occasional inconvenience.

Nor has age-gating remained confined to pornography or material encouraging suicide and self-harm. Big Brother Watch records restrictions on Goya’s Saturn Devouring His Son, discussion of Roman and Crusader history, news reporting from Gaza, footage from Ukraine, music videos and political speech. TikTok briefly removed a Reform UK video about immigration, despite the Act’s supposed protection for content of democratic importance. X restricted a video in which the Conservative MP Neil O’Brien discussed Britain’s falling birth rate. The video included a baby, which may now be an adult theme.

The report’s investigation of Reddit is particularly effective. Big Brother Watch examined eight large, ostensibly non-adult discussion groups and found ordinary posts caught behind the platform’s age barrier. These included a request for advice about assembling a care package for a friend undergoing prostate-cancer treatment, a complaint about abandoned bags of dog waste, a discussion about helping a relative with alcohol dependency, the poster for a mainstream horror film, music by the Moody Blues, and a photograph of a paper cut.

One restricted post showed milk escaping from a damaged bottle. Reddit users contributed the inevitable joke about “Hole Milk”. A system of British Internet regulation promoted as the answer to grooming, suicide and pornography had culminated in adults proving their age to look at a leaking bottle of milk. This is funny only because the alternative is despair.

The report understands that the apparently trivial examples disclose a serious mechanism:

“The substantial penalties available under the OSA create a strong incentive for platforms to be conservative and err on the side of caution when applying age restrictions to speech” (p.41).

That sentence contains the central truth about the Act. Parliament did not need to order Reddit to hide paper cuts or X to restrict a political discussion about birth rates. It needed only to threaten companies with large penalties and leave them to calculate the safest response. A platform gains little by defending a doubtful post and may lose greatly by allowing access. It will therefore remove or restrict more than the law expressly demands.

The censorship is delegated, but the coercion remains public. Ministers can deny responsibility because a private company pressed the button. The company can blame the regulatory environment. Ofcom can point to duties created by Parliament. Everyone has participated, and no one is responsible. This is the ideal arrangement for the modern managerial State.

Big Brother Watch sensibly does not pretend that Reddit bears no responsibility. Its classifications are imprecise. A post can inherit a “not safe for work” label from the history of the user who submitted it, and a tag originally intended to warn people about opening something in an office has been repurposed as the gateway to legal adulthood. But the report correctly identifies the root cause: Reddit has a legal duty to restrict certain categories and a powerful incentive to use its existing, broader classification system. The absurdity is not evidence that the law is innocent. It is evidence of how predictable commercial behaviour interacts with badly designed coercion.

More troubling than the jokes and cultural material are the discussions in which people seek help. Age-gating caught users looking for information about domestic violence, HIV exposure, sexuality, depression and medical treatment. A system advertised as protection may deny young people access to precisely the conversations in which they can ask for assistance. The Internet before regulation contained danger. The regulated Internet also contains danger, together with a digital bouncer demanding papers from those trying to reach a support group.

The policy creates a union between censorship and surveillance. A person may retain his privacy if he accepts restricted access, or gain access if he surrenders information about himself. The old censor was content to remove a book from the shelf. The new system wishes to know who is asking for it. Britain is creating what the report calls a “Great British Firewall, erected brick by brick through successive restrictions rather than a single legislative act” (p.41).

The next section turns from regulatory censorship to criminal punishment. Section 1 of the Malicious Communications Act 1988 and section 127 of the Communications Act 2003 contain broad prohibitions on indecent, grossly offensive or menacing messages. The Online Safety Act has added an offence concerning false communications sent with the intention of causing non-trivial harm. These laws regulate millions of spontaneous remarks through concepts whose application depends heavily on context and official judgment.

The report begins this discussion with Lord Justice Sedley’s famous statement in Redmond-Bate v Director of Public Prosecutions:

“Free speech includes not only the inoffensive but the irritating, the contentious, the eccentric, the heretical, the unwelcome and the provocative provided it does not tend to provoke violence. Freedom only to speak inoffensively is not worth having” (p.44).

This ought to be printed above the entrance of every police station. It might usefully replace some of the corporate declarations about values, community engagement and inclusion. The police do not exist to ensure that public expression is agreeable. If a statement is threatening, fraudulent or part of a campaign of harassment, the criminal law may properly intervene. Mere offence is not enough. In a country of nearly seventy million people, everything worth saying will offend someone.

Big Brother Watch obtained figures from police forces through Freedom of Information requests. Between 2021 and 2025, at least 62,199 people were arrested for the communications offences examined. The report estimates that the full total may have been about 73,000 because several forces supplied incomplete figures or refused the requests on cost grounds. The recorded figures include 18,520 charges and 12,292 convictions.

These categories should be compared with care. Not every arrest without a charge was necessarily improper, and the conviction figures do not cover the whole United Kingdom on an exactly equivalent basis. The report acknowledges the gaps in its data. Its strongest finding is not a simple calculation of guilt and innocence, but the extraordinary variation between police areas.

Cumbria Constabulary recorded 25.7 arrests per 10,000 residents during the five-year period. Staffordshire recorded 0.6. Surrey and Northumbria were also near the bottom of the table, while Northamptonshire and Essex were near the top. The report dryly observes that it “cannot be argued that people in Cumbria are orders of magnitude more offensive than their Northumbrian neighbours” (p.11).

The inhabitants of Cumbria may have robust opinions, but it is difficult to believe they are more than forty times as criminally communicative as the inhabitants of Staffordshire. The figures point instead to a postcode lottery created by vague legislation and local police culture. Whether words lead to an arrest may depend less on what was said than on which constabulary receives the complaint.

The ratio of arrests to charges varies almost as dramatically. Some forces appear to use arrest sparingly; others use it as an opening move. This matters because arrest is itself an interference with liberty. The suspect may be detained, searched and questioned. His devices may be seized. Bail conditions may restrict his movements or expression. He may lose work, acquire an enduring Internet reputation as the object of a police investigation, and spend months waiting for a decision. A later announcement that no further action will be taken does not return the lost time or remove the fear.

The case studies give life to the figures. Ian Johnson, a 68-year-old man whose dog had been fatally injured, was prosecuted after using abusive language during a bitter dispute. He was eventually acquitted, but only after two years of legal trouble. A vulnerable teenage girl in the West Midlands was threatened with arrest and had her devices demanded because of an insulting TikTok post about a teacher, although she had apparently done no more than view it. Basic inquiries later showed that she had not created or distributed the material. Police nevertheless published a statement that, according to her mother, made her identifiable and worsened the harm done to her.

Most grotesque is the case of wildlife campaigner Anneka Svenska. Police accused her of knowingly publishing false information about the disposal of animals bred for research and invited her to accept a “community resolution”. This included an online “thinking skills” course. She supplied documents supporting her claim, whereupon the investigation was reportedly dropped within fifteen minutes.

A State that sends officers to improve a campaigner’s thinking skills has left satire unemployed. Yet the phrase is useful. It states openly what much of this machinery assumes: controversial speech is evidence of defective cognition, and official intervention is a form of correction. The dissident no longer requires punishment. She requires training.

The report also contrasts police enthusiasm for communications arrests with the use of voluntary interviews in some investigations into online child sexual abuse. Resource management was reportedly one reason for avoiding arrest in those graver cases. The comparison is devastating. Police forces that claim to lack the resources to arrest suspected sexual offenders can somehow find two officers and an afternoon when a campaigner may need her thinking improved.

The protest section demonstrates that neither side of the conventional political division can claim clean hands. The Police, Crime, Sentencing and Courts Act 2022 and the Public Order Act 2023 were Conservative measures. They expanded police authority to impose conditions on demonstrations, created new protest offences and introduced suspicionless stop-and-search powers. The legislation reached static assemblies, noisy protests and even one-person demonstrations. Everyday objects could become evidence of preparation for “locking on”. Civil Serious Disruption Prevention Orders can restrict association, movement and Internet use, with imprisonment available for breach.

These powers were created in response to unpopular campaigners, particularly environmental protesters. Conservatives who disliked Extinction Rebellion were willing to enlarge the permanent authority of the State in order to suppress a temporary nuisance. They might have reflected that powers do not retain the political opinions of the ministers who create them.

The Conservative defence is that disruptive protest interferes with the rights of others. This is true. Blocking an ambulance or preventing a person from reaching his work cannot be dismissed as mere expression. But laws against obstruction, trespass, criminal damage and public nuisance already existed. The new legislation was designed not merely to punish specific wrongdoing, but to give the authorities wider control over who might protest, where, with what objects and under what conditions.

Suella Braverman used secondary legislation to define “serious disruption” as anything “more than minor”, although Parliament had rejected that standard. The courts overturned her regulations. This episode gives us Conservatism in its modern form: a minister claiming to defend the constitution while attempting to revive through executive regulation a provision that the legislature would not enact.

Labour has not restored the presumption of liberty. The Crime and Policing Act 2026 introduced a new offence of concealing one’s identity in an area designated by police. Previously, an officer could require the removal of a face covering in specified circumstances. Now wearing one in a designated area may itself lead to imprisonment or a fine. This is especially revealing when the same political order is promoting digital identification online. The citizen is increasingly expected to identify himself both when he speaks through a computer and when he stands in a crowd.

The report examines restrictions imposed on demonstrations from very different parts of the political spectrum. Pro-Palestine marches have been moved and conditioned. UKIP assemblies have also been displaced. Suspicionless searches appear to have been used principally at one pro-Palestine protest, but the existence of the power threatens everyone. The report’s non-partisanship is one of its greatest strengths. It defends people carrying “Globalise the intifada” placards, a man who displayed a crudely worded attack on Islam, and anti-monarchists detained during royal events.

This is what belief in freedom of speech looks like. It does not mean deciding that every disputed slogan is wise, humane or historically accurate. “From the river to the sea” and “globalise the intifada” have meanings that are genuinely contested and may be heard as threats by people with good reason to fear political violence. But ambiguity is an argument against automatic arrest, not for it. Context, intention and the likelihood of actual harm matter.

Peter Tatchell was arrested while carrying a sign that explicitly defined his appeal to intifada as “Non-violent resistance”. Proceedings were dropped, but not before bail conditions had kept him away from Palestine-related protests for three months. Police had initially threatened him with a complete ban on entering central London and the confiscation of his telephone. A man was therefore removed from political activity not by conviction after a trial, but by conditions imposed while the authorities considered whether his explicitly non-violent placard might be unlawful.

At the coronation, members of the anti-monarchy group Republic were arrested and held for up to sixteen hours after police mistook luggage straps used to secure placards for devices intended for locking on. The Metropolitan Police later expressed regret. In Oxford, Symon Hill was arrested after shouting “Who elected him?” during the proclamation of Charles III. The charge was dropped, and Thames Valley Police eventually paid compensation.

The monarchy survived the question. A political order confident in its legitimacy ought to survive a placard, a chant and an impertinent voice in a crowd. The spectacle of police protecting public institutions from mild insolence does more damage to their dignity than the insolence itself.

The final substantive section concerns the National Security Online Information Team, previously called the Counter Disinformation Unit. Its declared purpose includes identifying foreign information operations, which is a legitimate concern. Hostile governments do conduct propaganda and deception. What does not follow is that Whitehall should monitor domestic political narratives, pass material to social-media companies and conceal the details behind claims of national security.

Big Brother Watch shows how the distinction has blurred. During the disturbances following the Southport murders in 2024, the unit monitored discussion of immigration and “two-tier policing”. Officials flagged “concerning narratives” to TikTok and asked what action had been taken. Whether two-tier policing exists is a matter for evidence and argument. It is not transformed into foreign disinformation because ministers dislike the allegation.

The unit has worked with an artificial-intelligence platform developed under a £2.3 million contract. Technology derived from tools for tracking terrorism can be redirected towards other “priority areas”. The report is careful not to claim that all monitoring is censorship. Governments and other large organisations will inevitably observe public discussion. The danger arises when a government body with direct access to platform enforcement teams identifies domestic opinions as a security problem and refuses to disclose how it operates.

The David Cameron prank-call episode is almost too perfect. Two Russian pranksters deceived the then Foreign Secretary into speaking with them and published the resulting conversation. They openly presented it as a prank. A government contractor nevertheless produced an assessment, much of which was withheld on national-security grounds, although the disclosed portions apparently found no manipulation of the recording.

As the report observes, “Despite being a hoax call that embarrassed the government and Lord Cameron, it is difficult to see how an overt prank fits into the definition of disinformation” (p.100). There lies the danger in miniature. The State’s counter-disinformation machinery had turned its attention to truthful material obtained by deception because the truth embarrassed a minister. “Disinformation” was ceasing to describe the falsity of a claim and beginning to describe the inconvenience of its circulation.

Here Labour’s responsibility is direct. It has continued the unit, developed its technical machinery and entertained further proposals for managing the information environment. Sadiq Khan has called for a more powerful central body to combat disinformation. Government proposals have included giving officially “trustworthy” media greater prominence during crises. Labour may speak the therapeutic language of cohesion and resilience where Conservatives preferred safety and order, but the institutional appetite is the same. Ministers wish to influence what is visible, decide which sources deserve elevation, and monitor the opinions that their policies provoke.

The Conservatives built much of the apparatus described in this report. Labour has retained it and, in several respects, enlarged it. This is not inconsistency between the parties. It is continuity within the State. Each party denounces a particular use of coercive power while in opposition, then discovers the general usefulness of that power after entering office. The machine does not care which rosette is pinned to its operator.

I have two reservations about the report, neither of which alters my strongly favourable judgment. The first is that its principal recommendation—an independent national review of the laws affecting speech—is too modest. Reviews are among the preferred methods by which the British State converts outrage into stationery. A review may be valuable if it identifies provisions for repeal, but the evidence already assembled here justifies immediate action. Vague communications offences should be rewritten or abolished. Suspicionless protest searches should go. Government contact with platforms concerning lawful domestic political speech should be disclosed and tightly limited. The Online Safety Act requires more than adjustment around the edges.

My second reservation is that the report occasionally gives legislative intention more respect than it deserves. It repeatedly distinguishes intended restrictions from unintended consequences. This is fair to individual legislators, many of whom probably did not wish to age-gate paper cuts or close a forum for fathers. But Parliament deliberately constructed a system in which severe penalties encourage private companies to suppress doubtful material. Over-removal was not an unforeseeable accident. It was built into the incentives.

Mrs Dorries’s foreword embodies this weakness. She asks us to distinguish the pure intention from the dirty result, and then to regard the sponsors of the policy as fellow mourners at its funeral. No. A person who pours petrol across the floor does not escape responsibility because someone else struck the match. Nor does she become a fire-safety expert by expressing regret beside the ashes.

These are arguments for going further, not reasons to discount the report. Big Brother Watch has performed an important public service. It combines legal analysis, official statistics obtained by persistent inquiry, direct testimony and examples from across political boundaries. It neither denies real harms nor accepts that invoking them ends the argument. It understands that child exploitation, foreign propaganda, threatening communications and serious disruption exist. It asks the question politicians prefer to avoid: whether the powers created in response are necessary, proportionate and subject to meaningful control.

The answer is frequently No.

This report should be read by MPs, journalists, police leaders and anyone who still imagines that censorship begins only when Parliament expressly bans an opinion. It should also be read by campaigners who defend speech only when they approve of the speaker. Liberty divided by faction is not liberty. It is a competition for possession of the censor.

Britain’s free-speech problem was not imposed in one coup and will not be cured by replacing one party with the other. It has grown through the accumulated work of Conservative and Labour governments, assisted by frightened legislators, ambitious officials and companies responding rationally to official threats. Every new restriction has been another brick in the Great British Firewall.

The wall is not yet complete. This report has done us the service of showing where the bricks were laid and who supplied them. Nadine Dorries has done us the lesser service of returning to the scene and admitting that she was one of the builders.