Where Does Freedom of Speech End? — Secrets, Propaganda and Political Speech
How do Western democracies balance speech, secrecy, and counter-propaganda? We examine whistleblowers, state media, protests, and official restrictions.
AI-generated symbolic editorial illustration, not a documentary image.
Freedom of speech is one of the values Western democracies most often present as part of their political identity. Yet almost none of them treats that freedom as absolute.
Article 10 of the European Convention on Human Rights protects the right to receive and impart information and ideas, while expressly permitting restrictions prescribed by law and necessary in a democratic society for reasons including national security, public safety, prevention of crime, protection of the rights of others, or preventing the disclosure of confidential information. So the question is not whether a state may ever restrict speech.
The question is how quickly speech that troubles those in power becomes a “state secret”, “propaganda”, “disinformation”, “foreign interference” or “dangerous disruption of public order” — and whether the West would accept an equally broad interpretation of those concepts if it were used by a state it criticises. This article therefore deliberately separates three different things.
Disclosure of classified state information is not the same as the operation of a foreign state-linked media outlet.
A foreign state-linked media outlet is not the same as a domestic protester. And a private social-media company’s moderation decision is not the same as state censorship. If we collapse those categories, we get propaganda instead of analysis. If we keep them separate, the double standard becomes much easier to see.
When a state secret collides with the public interest
A state needs secrets. Intelligence sources, military plans, agent identities, active operations and technical capabilities can endanger lives and national security if disclosed recklessly.
But the label “classified” does not automatically mean “morally justified in being hidden from the public”.
The history of Western democracies itself contains cases in which confidential documents exposed state conduct that the public might otherwise never have known.
That creates a conflict that cannot be resolved by slogan. The state sees a breach of the duty to protect secrets. The public may see the same act as exposure of government conduct. Both claims can be partly true at the same time.
Snowden: exposure of surveillance and criminal prosecution existed at the same time
In 2013, Edward Snowden provided journalists with a large body of classified NSA material. The disclosures triggered a global debate about mass surveillance, Section 215, communications-data collection and the relationship between security and privacy.
But the U.S. legal response was not a determination of whether the disclosures were in the public interest.
In June 2013, the Department of Justice charged Snowden, among other things, under 18 U.S.C. § 793(d) for unauthorised communication of national-defence information and under 18 U.S.C. § 798 for disclosure of classified communications-intelligence information. [1]
The texts of 18 U.S.C. § 793 and § 798 criminalise certain forms of unauthorised communication or disclosure of national-defence information and classified communications-intelligence information. Those provisions contain no general express exception allowing public disclosure to the media merely because the person disclosing the information believes that doing so serves the public interest. [2] [3]
At the same time, U.S. law provides protected channels for reporting serious wrongdoing inside the intelligence community. Under 5 U.S.C. § 416, an employee may report an urgent concern to the appropriate Inspector General and, under specified conditions, to the congressional intelligence committees, with security procedures applying where classified information is involved. Such a protected disclosure is not the same thing as publicly providing classified material to a media organisation. [4]
That is an important legal distinction. The question whether the state was operating an excessive or legally questionable surveillance programme is not the same question as whether an individual had the legal right to remove classified documents and provide them to journalists. But precisely because of that separation, a democratic paradox can emerge:
a disclosure may help trigger reform of a state programme, while the person who enabled the disclosure is still treated primarily as someone who violated secrecy laws.
That is not necessarily a legal contradiction. But it is a political and moral question that neither the slogan “whistleblower” nor the slogan “traitor” can settle by itself.
Hale and Winner: motive does not erase the act
Daniel Hale and Reality Winner further show why these cases should not be romanticised. Hale admitted that, as an intelligence analyst and later contractor, he obtained and provided a journalist with classified national-defence information. In 2021, he was sentenced to 45 months in prison. [5]
Reality Winner printed a classified intelligence document, removed it from a secure environment and sent it to a news organisation. In 2018, she pleaded guilty to unlawful retention and transmission of national-defence information and received a 63-month prison sentence. [6]
In both cases, the state had a legitimate interest: people entrusted with access to classified information cannot unilaterally decide that every document they regard as important ceases to be classified. But the opposite extreme is dangerous too.
If classification itself ends the debate about public interest, the executive branch holds enormous power over which facts about its own conduct can ever become part of democratic scrutiny. The harder question is therefore:
are the lawful internal channels independent, safe and effective enough for a person to report serious wrongdoing when the evidence itself is classified — without being forced to choose between silence and unauthorised public disclosure?
The existence of statutory routes to Inspectors General and congressional oversight is an important corrective. By itself, however, it does not answer how effectively those routes work in any particular case.
Assange: a case that cannot be reduced to “a journalist imprisoned for publishing”
The case of Julian Assange is even more sensitive. His prosecution became a symbol of the danger that the Espionage Act could be used against publication of classified material. But the legal case was not limited to passive receipt and publication of documents.
In June 2024, Assange pleaded guilty to conspiring with Chelsea Manning to unlawfully obtain and disclose classified national-defence information. Under the plea agreement, he received a sentence of time already served — 62 months — and returned to Australia. [7]
It would therefore be inaccurate to say that the U.S. government simply convicted an ordinary newspaper publisher solely for publishing truthful information. But the danger the case raised for journalism is not imaginary either.
When criminal law designed to protect national-defence information is used in a case involving the acquisition and mass publication of government documents, the boundary between criminal procurement of secrets and protected journalistic activity becomes critically important. A democracy needs that boundary to be clear.
The Pentagon Papers: the key Western counterargument
That is why New York Times Co. v. United States from 1971 matters so much. The U.S. government tried to stop The New York Times and The Washington Post from publishing a classified study of U.S. decision-making on Vietnam — the Pentagon Papers.
The Supreme Court refused the requested prior restraint. In its per curiam decision, the Court held that the government had not met the “heavy burden” required to justify such a prior restraint.
The case became one of the strongest symbols of American protection of press freedom. Official decision: [8] But here too the comparison must remain precise.
The Pentagon Papers did not establish a general rule that every government employee has a right to disclose every classified piece of information.
The Court was addressing the extraordinarily heavy burden the government must satisfy if it wants to prevent a newspaper from publishing in advance.
That is a different legal situation from criminal prosecution of a person who unlawfully obtained or disclosed the material.
The distinction can be frustrating. But it matters. The Western system therefore contains two principles at once: the state may punish certain unauthorised disclosures of classified information;
and the state faces an extraordinarily high burden if it seeks to stop a media organisation from publishing in advance. That is real protection of press freedom.
It also means that the person who gets the document to the press may be far more legally exposed than the institution that ultimately publishes it.
Who decides when secrecy matters more than disclosure?
The system has a logic. The journalist did not sign a security obligation; the intelligence officer did.
The journalist’s role is to inform the public; the official’s role includes protecting classified information. But the universalist test remains.
When Western states condemn the prosecution of whistleblowers, dissidents or journalists elsewhere, they often emphasise the public interest in disclosure.
When a disclosure concerns their own secrets, the weight can shift quickly toward secrecy obligations and national security. Sometimes that shift is entirely justified. But not automatically.
Their security justification should be understood — not adopted as an acquittal.
A second layer: a foreign state-linked outlet is not an ordinary whistleblower
The next category is fundamentally different. RT and Sputnik are not individuals who removed a document and handed it to a journalist. They are media systems linked to the Russian state and aimed at international audiences. They should therefore not be analysed under the same test as Snowden or Hale. And it makes even less sense to say: “RT receives state funding; the BBC or Voice of America does too; therefore they are all the same.” They are not.
State funding alone does not determine the degree of editorial control.
In 2017, the U.S. Department of Justice required the production company T&R Productions to register under the Foreign Agents Registration Act as an agent of ANO TV-Novosti, the Russian government entity responsible for broadcasting the RT network. [9]
Voice of America is also publicly funded, but U.S. law requires respect for its professional independence and integrity, and its institutional model includes a formal editorial firewall. [10]
That does not prove that VOA has been perfectly politically neutral in every historical circumstance. It does mean that the equation “public money = the same kind of state media” does not hold.
The relevant comparison is the actual legal status, editorial safeguards, relationship to the state, content control and operational role.
The EU did in fact remove RT and Sputnik from the distribution space
In March 2022, the European Union suspended the broadcasting activities of RT and Sputnik in or directed at the EU.
The Council justified the measure on the ground that Russia was systematically using information manipulation and disinformation as an operational instrument linked to military operations in Ukraine and that this posed a threat to public order and security in the Union. [11]
The measure did not remain limited to the original two systems. The EU later expanded the list. The Council’s current explanatory material identifies 27 Kremlin-backed media organisations whose broadcasting activities or licences have been suspended. [12]
This is a real restriction on media distribution. It is not merely a label, warning or fact-check. That makes the question legitimate:
how does a political union that presents freedom to receive information as a fundamental right justify preventing a specific foreign media outlet from broadcasting?
The EU has an answer.
RT France lost in court as well
RT France challenged the measure before the General Court of the European Union. In July 2022, the Grand Chamber of the General Court dismissed its action.
The Court accepted that the Council has broad discretion in the field of the Common Foreign and Security Policy and that, in the exceptional wartime context, it could temporarily prohibit the broadcasting activities of a media outlet funded by the Russian state budget where that outlet supported Russia’s military action.
The Court also examined freedom of expression and information and regarded the measure as proportionate in the circumstances. [13]
RT France appealed, but appeal proceedings in Case C-620/22 P ended in July 2023 when the case was removed from the register; the General Court judgment was therefore not overturned on appeal. [14]
This is an important corrective. It would not be fair to say that the EU simply silenced RT without legal process or access to judicial review.
Judicial review existed. RT France used it. And it did not succeed in having the measure set aside. But that is precisely why the case matters for this article.
The EU legally demonstrated that even a liberal freedom-of-expression framework permits a very strong restriction on a foreign media organisation where the state or political union considers it part of a hostile information operation.
“Propaganda” by itself is not a precise enough category
This is where the problem of language begins. The word propaganda can quickly move from description to permission. Once something is labelled propaganda, restricting its distribution becomes psychologically much easier. But a great deal of political communication is propagandistic in a broad sense. States finance strategic communication. Militaries conduct information operations. Governments select data, highlight successes and minimise failures. Political campaigns manipulate emotion. Media organisations have editorial perspectives. A democratic boundary therefore cannot simply be: “their speech is biased.” A more precise standard is needed. That is where the concept of Foreign Information Manipulation and Interference — FIMI becomes useful.
FIMI: the EU itself acknowledges that much of the problem is not illegal
The European External Action Service describes FIMI as a form of manipulative, intentional and coordinated behaviour that is mostly non-illegal yet can negatively affect values, procedures and political processes.
The EEAS specifically monitors Russia and China, as well as state and non-state actors, proxies, diplomatic networks, state media, social platforms and other channels. [15]
That admission is central to the double-standard test. The problem of foreign influence is not limited to falsehoods.
It can include coordination, covert amplification, false identities, manipulation of distribution or strategic exploitation of truthful information. But the implication is sensitive:
if FIMI is often legal speech or legal conduct, then countering FIMI cannot rest only on the legality of content.
The state begins to assess intent, coordination, origin and effect. That may be necessary to protect elections and democratic processes.
But it is also an area in which a security category can begin to capture political speech that is unpleasant yet lawful.
The American model showed a milder alternative: disclosure of origin
Before the European ban on RT, the American approach offered a different model. In 2017, the Department of Justice required RT’s U.S. production company to register under FARA. The logic was not: “Americans must not watch this.” It was:
the public should know who is acting on behalf of a foreign principal.
FARA has its own controversies and limitations, but the model illustrates an important difference between: disclosure of origin, labelling, restriction of reach, and full suspension of distribution. A democracy does not have only the choice between complete passivity and a ban. It has a spectrum of measures. That is why, when choosing the strongest measure, it should show why a milder one is insufficient.
A third layer: domestic political speech
Uradni dokument — www.govinfo.gov
The most sensitive test comes when the speaker is no longer a foreign media organisation or a person handling state secrets.
When the speakers are the state’s own citizens. The same foreign-influence argument can no longer be used. But domestic speech is not absolute either. A protest can block a road. A demonstration can disrupt infrastructure. A threat is not protected in the same way as a political opinion. The state has obligations to protect other people too. The question is again proportionality.
Britain: the right to protest and “serious disruption”
The United Kingdom has significantly expanded legal tools against especially disruptive forms of protest in recent years.
The Public Order Act 2023 introduced or regulated, among other things, new offences and Serious Disruption Prevention Orders.
Under the current Section 21, a court may make such an order on application by the police against an adult where, among other conditions, the person has at least two relevant protest-related convictions or contempt findings within the specified period and the court considers the order necessary to prevent further serious disruption. [16]
The government had an intelligible justification. Groups such as Just Stop Oil and Insulate Britain used tactics designed to maximise disruption to roads and infrastructure. A right to protest is not a right without limits, and the rights of other people also matter.
But even before the bill became law, the British Joint Committee on Human Rights warned that the proposed package risked creating a chilling effect on peaceful protest and upsetting the balance between protesters’ rights and the rights of the public. More strikingly, the Committee framed the problem in foreign-policy terms as well: if the state itself adopts overly broad restrictions on peaceful protest, it weakens its credibility when criticising other countries for similar conduct. [17]
That is close to a textbook example of the problem examined in this series. The criticism did not come from Moscow or Beijing. It came from the British Parliament.
The government’s response does not resolve the breadth of the powers
The British government rejected the Committee’s assessment. It argued that the right to peaceful protest does not include a right to criminality, disorder or disproportionate interference with the rights of others and that additional powers were needed because of repeated forms of deliberate, serious disruption of infrastructure. [18]
The formal breadth of the power has to be separated from how often it has actually been used. In a post-legislative review published in June 2026, the Home Office reported that SDPO powers had been in force since 5 April 2024, but use through the end of 2025 was very limited. Of 40 police forces that returned data, 33 reported no applications or adjournments relating to SDPOs, while the remaining seven were unable to provide data. HM Courts & Tribunals Service said volumes were so low that even summary figures could not be published without risking identification of individuals. The 40 responding forces also reported no arrests or charges for breaches of SDPOs. [19]
A broad legal instrument and the frequency with which it is actually used are not the same thing. Very limited use through the end of 2025 weakens claims of mass practical deployment, but it does not answer why such a broad preventive power was enacted in the first place. But the same rule still applies:
the justification should be understood, not automatically accepted as an excuse.
If another state were to restrict the movement of known protesters preventively, expand police powers and justify those measures by public order and protection of infrastructure, Western institutions would ask: is the measure necessary? is it proportionate? is it sufficiently precise? is there independent oversight? is it used against violence, or also against politically inconvenient expression? The same questions have to apply at home.
Social media: private moderation or state censorship?
Digital platforms create an even harder boundary. Facebook, YouTube, X and other platforms are private companies. They may create their own rules, remove content and restrict accounts within the limits of law.
The First Amendment to the U.S. Constitution generally constrains the state, not a private editor or platform.
But what happens when the government begins telling a platform which content it considers false, dangerous or manipulative?
The fact of communication alone is not censorship. A government may warn a company about a terrorist account, a foreign influence operation, a cyber threat or health misinformation. The boundary appears when persuasion becomes coercion.
The Supreme Court: the state may persuade, but may not coerce a private party into suppressing speech
In NRA v. Vullo, the U.S. Supreme Court in 2024 reaffirmed an important principle: a government official may not directly or indirectly coerce a private party into punishing or suppressing speech the government disfavors.
The key question is whether, in context, the government communication operates as persuasion or as a threat of adverse government action. [20]
That is a strong protection. But another case from the same year showed how difficult that boundary can be to prove in a large digital system.
Murthy v. Missouri: the Court did not rule that there had been no censorship
In Murthy v. Missouri, the plaintiffs argued that federal officials pressured social-media companies to restrict certain content, particularly concerning Covid-19 and elections.
In 2024, the Supreme Court reversed the lower court’s decision and denied the requested injunctive relief. But the reason matters. The majority did not announce a general holding that: “the federal government did not pressure platforms.”
It held that the plaintiffs had not established the Article III standing required for the prospective injunction they sought. It emphasised the need for a concrete link between a particular government official, a particular platform, a particular moderation decision and harm to a particular plaintiff. The majority also noted that platforms had independent incentives to moderate and often exercised their own judgment. [21]
Justices Alito, Thomas and Gorsuch read the record very differently in dissent. In their view, the circumstances satisfied the factors the Court used in Vullo to distinguish permissible persuasion from impermissible coercion, and the evidence showed coercion.
The same official source contains both the majority and dissenting opinions: [21]
It is therefore inaccurate to use Murthy as proof that the entire story of government pressure was invented.
It is equally inaccurate to say that the Supreme Court found a broad federal censorship system unconstitutional. It did not. It did not resolve the merits in a way that permits that categorical conclusion.
In 2025, the U.S. executive branch changed direction itself
On 20 January 2025, the new U.S. administration issued Executive Order 14149, Restoring Freedom of Speech and Ending Federal Censorship.
The order declares a policy that federal officials may not use government resources to unconstitutionally restrict protected speech and directs the Attorney General to review federal activities from the previous four years. [22]
Importantly: the language of the executive order reflects the political and executive assessment of the new administration.
It is not a judicial finding that all the conduct it describes by the previous administration was unconstitutional.
But the policy shift itself confirms that the line between lawful government warnings to platforms and impermissible pressure is not an academic detail. It became an issue of state policy.
“Disinformation” can be a description — or an instrument of power
The hardest question shared by all three layers is the same. The state has legal powers in defined circumstances to protect secrets, respond to covert foreign influence operations and maintain public order. In some circumstances it may also alert platforms to dangerous or manipulative activity. But every one of those powers contains the same temptation:
the state itself defines the category that allows it to treat speech more harshly.
Secret. Propaganda. FIMI. Disinformation. Serious disruption. Security threat. Those categories are not invented.
The problem begins when a descriptive concept turns into a shortcut around the tests of necessity and proportionality.
The real test is not: “does the West have a right to restrict speech?”
Of course it does. No serious theory of free expression requires protection for handing operational secrets to a hostile intelligence service.
No democracy is required to tolerate a covert foreign influence operation. No right to protest includes an unlimited right to violence or destruction of infrastructure.
And no government official has the right to coerce a private intermediary into silencing lawful political speech merely because that speech is inconvenient to those in power. The real test is therefore different:
is the rule precise enough to restrict dangerous conduct — or broad enough to capture an unwanted opinion?
Judicial and parliamentary oversight do not erase the restrictions
On this subject, it would be especially unfair to ignore self-correction. The U.S. Supreme Court blocked prior restraint in the Pentagon Papers case. Courts can review measures against media organisations. RT France obtained judicial review, even though it lost.
A British parliamentary committee publicly warned its own government that protest restrictions threatened rights and British credibility.
The U.S. Supreme Court in Vullo clearly constrained government coercion of private intermediaries. In Murthy, the dispute over contacts between government and platforms reached the country’s highest court. These are real institutional differences. They do not prove that the West has no double standards. But they do show that it possesses tools capable of exposing and constraining them.
That is precisely why the double standard matters more, not less
If Western states openly said: “Freedom of speech is only our internal rule, and outside that we act according to interest,” the dispute would be simpler. But they do not say that. They present freedom of expression as a human right. Media freedom as a measure of democracy. Protection of journalists as an international norm. The right to peaceful protest as something other governments must respect too. And political pluralism as a universal value. By doing so, they choose the stricter standard themselves. So we have to ask: when another state closes a media outlet for “national security”, how much evidence do we demand? When it prosecutes someone for disclosing a state secret, do we ask what was revealed? When it restricts protest for “public order”, do we test proportionality?
When it pressures a platform over “disinformation”, do we call that protection of the information environment — or censorship?
If we ask those questions of others, we have to ask them of ourselves.
Freedom of speech is not demonstrated by speech we like
A democratic state does not prove its commitment to free speech by allowing speech that does not trouble it. The harder test comes with information that embarrasses it. With a media outlet it regards as hostile. With a protester who genuinely disrupts everyday life. With content the government regards as dangerous or false.
That is when we see whether freedom is a principle or merely the default condition until it becomes politically expensive.
This does not mean every restriction is wrong. It means every restriction must be demonstrably justified. Necessary. Proportionate. Limited in time and scope. Subject to independent oversight. And precise enough that protecting society does not become protecting the government from criticism.
Where, then, does freedom of speech end?
It does not end at one line. It ends differently for different kinds of speech. A state secret requires a different test from a newspaper. A covert foreign operation requires a different test from a publicly funded media outlet. A peaceful protest requires a different test from destruction of infrastructure. Private moderation requires a different test from state coercion. That is why it is dangerous to collapse everything under the word “censorship”.
But it is equally dangerous to wash away every state restriction with the words “national security”, “disinformation” or “public order”.
Western democracies have courts, parliamentary oversight, plural media and legal mechanisms that can force governments to explain their restrictions.
But the existence of those mechanisms does not answer whether a specific restriction was justified under the same standard the West demands from others.
Their universalist argument is convincing only if they apply those mechanisms to themselves with the same seriousness they demand from others.
Because freedom of speech is not proven when the government speaks. It is proven when government hears something it does not want to hear — and must still demonstrate why it has the right to silence it.
Sources and further reading
- US Department of Justice — request to Hong Kong for Edward Snowden’s provisional arrest Source
- US Code — 18 USC § 793, defence information Source
- US Code — 18 USC § 798, disclosure of classified information Source
- US Code — 5 USC § 416, inspectors general and intelligence activities Source
- US Department of Justice — sentencing of a former intelligence analyst for disclosing classified information Source
- US Department of Justice — sentencing of a contractor for transmitting classified material to a news outlet Source
- US Department of Justice — WikiLeaks founder’s guilty plea and sentencing Source
- US Supreme Court — New York Times Co. v. United States (Pentagon Papers), 403 US 713 Source
- US Department of Justice — production company registration under FARA Source
- US Agency for Global Media — editorial independence and the firewall Source
- Council of the EU — sanctions on RT and Sputnik broadcasting (2 March 2022) Source
- Council of the EU — explanation of sanctions against Russia Source
- EU General Court — judgment in RT France v. Council (2022) Source
- Court of Justice of the EU — appeal case C-620/22 P, RT France Source
- EEAS — information integrity and countering foreign information manipulation and interference Source
- United Kingdom — Public Order Act 2023, section 21 Source
- UK Parliament — human-rights scrutiny of the Public Order Bill Source
- UK Parliament — Joint Committee on Human Rights report on the Public Order Bill Source
- UK Government — post-legislative scrutiny of the Public Order Act 2023 Source
- US Supreme Court — National Rifle Association v. Vullo (2024) Source
- US Supreme Court — Murthy v. Missouri (2024) Source
- White House — executive order on freedom of speech and federal censorship (2025) Source