From Free Speech to Social Control: Australia’s Retreat from Liberty
For most of the history of Western democracy, freedom of speech rested upon a proposition that was simultaneously dangerous and magnificent: ordinary people should be permitted to hear competing ideas and decide for themselves what they believe. Governments might hate what was being said, majorities might find it offensive, while churches, corporations, universities and newspapers might regard particular opinions as foolish, dangerous or immoral. Nevertheless, the presumption was that the citizen, rather than the state, should ultimately decide what could be thought, heard and said. Australia is drifting away from that principle.
A recent essay by Ben Crocker of the Institute of Public Affairs places the present debate within the much longer Western history of free speech, a history running through democratic Athens, the constitutional restraints associated with Magna Carta, John Milton, the English Bill of Rights, the American founding and Australia's own liberal tradition. That history matters because freedom of speech did not simply appear one morning as an administrative policy that governments could adjust whenever circumstances changed. It emerged through centuries of struggle over who possessed the authority to decide what other people were permitted to say. The answer gradually reached by the Western tradition was revolutionary: government should possess much less of that authority than governments naturally desire.
That principle was never absolute. Ancient Athens could punish unacceptable speech, and Socrates discovered just how far Athenian tolerance extended when he was condemned to death in 399 BC. Yet Athens also developed the concepts of isegoria, the equal right to participate in public speech, and parrhesia, the freedom to speak frankly. The importance of Athens was not that it created some modern libertarian paradise, because plainly it did not, but that open public argument became central to political life.
The development thereafter was long, uneven and sometimes contradictory. Magna Carta was not a free-speech charter, but it became part of the wider constitutional tradition that rulers themselves were subject to law. Milton's Areopagitica (1644) attacked the licensing of printing and made one of the enduring arguments against prior restraint, even though Milton himself did not advocate an unlimited modern conception of free expression. The English Bill of Rights protected freedom of speech in parliamentary debate, while Madison and the American founders went further by attempting to place important expressive liberties beyond ordinary governmental interference.
Running through these developments was a simple but profound insight: rulers cannot safely be trusted to determine which opinions the population should be permitted to encounter. That lesson seems increasingly uncomfortable for Australia's political class.
The modern justification for restricting speech is rarely presented as censorship. Nobody establishes a Ministry of Truth and hangs a sign above the entrance. The language is therapeutic and managerial. Governments speak of "harm," "safety," "misinformation," "disinformation," "online wellbeing," "duty of care" and protecting vulnerable communities. Each expression sounds reasonable in isolation, but the difficulty begins when government acquires the authority to determine how broad and elastic concepts of harm should operate in political and social communication.
The Albanese Government's proposed Digital Duty of Care illustrates the change particularly well. On 8 September 2026 the government released the exposure draft of the Online Safety Amendment (Digital Duty of Care) Bill 2026, building upon its commitment to require online services to take reasonable steps to prevent foreseeable online harms. The scheme moves beyond merely responding to clearly unlawful material after harm has occurred and towards requiring platforms to identify and mitigate risks in advance, including risks associated with categories of "seriously harmful material."
The government's argument is deceptively straightforward. Large digital platforms design the systems through which harmful material can be amplified to millions of people, so responsibility should not fall entirely upon individual users after the damage has already occurred. Platforms should be required to consider the risks created by their own systems and take reasonable preventive measures.
The free-speech concern appears when regulation moves beyond clearly unlawful conduct into broader and more contestable categories of harm. The more elastic the category, the more its boundaries depend upon statutory definitions, regulatory judgement, administrative guidance and corporate risk management. At that point regulation begins moving from the relatively concrete question of whether particular conduct is illegal towards the much wider question of whether allowing particular forms of communication creates an unacceptable social risk.
Traditional liberalism asked whether the state possessed sufficient justification to prevent a citizen from speaking. The emerging administrative philosophy increasingly asks whether permitting citizens to speak without supervision creates an unacceptable risk. Those questions may look similar, but they begin from opposite presumptions. Under the first, liberty is the default and restriction requires justification. Under the second, expression increasingly becomes an activity whose consequences are to be assessed, managed and mitigated.
The citizen consequently changes status. He is no longer simply an autonomous adult entitled to encounter good arguments, bad arguments, lies, truths, eccentricities, propaganda and offensive opinions. He increasingly becomes a potential victim of harmful information whose informational environment must be made safer for him. That is a significant philosophical change, because there is no neutral instrument capable of determining what constitutes harmful political speech.
Someone must exercise judgement. Legislators establish broad categories, regulators develop standards and guidance, bureaucrats interpret them, lawyers advise companies about regulatory exposure and platforms construct automated systems designed to minimise commercial and legal risk. At every stage human beings exercise judgement, and human beings inevitably bring assumptions, values and political perspectives to those judgements.
The result need not involve a government official telephoning a newspaper editor and ordering an article removed. Twenty-first-century censorship can operate much more indirectly. Government establishes sufficiently broad regulatory obligations, platforms become legally and commercially risk-averse, and private companies remove, downgrade or suppress material before government ever needs to issue an individual command. The individual censor can therefore become increasingly difficult to identify even while the machinery capable of restricting speech expands.
This is why regulation through duties of care deserves particular scrutiny. If a platform faces substantial consequences for failing to control material later judged to create unacceptable harm, while facing little comparable penalty for unnecessarily removing legitimate material, its incentives will naturally favour caution. When uncertainty arises, the commercially rational response may be to restrict distribution rather than risk regulatory action.
Government can then maintain, perhaps correctly in a formal sense, that it did not censor anybody because the individual moderation decision was made by the platform. The platform can reply that it merely complied with its legal obligations and managed foreseeable risks. Each actor can plausibly deny having deliberately suppressed a political opinion, while the cumulative structure nevertheless produces less space for controversial expression.
This represents a broader transformation in political culture. Governments have always regulated conduct, but contemporary governance increasingly extends into the informational and psychological environments within which citizens make choices. The object is not merely to prohibit defined wrongdoing but to identify risks in advance and reduce exposure to them. That philosophy may be defensible in some areas, particularly where children, fraud, exploitation or direct criminal conduct are concerned, but it becomes much more controversial when applied to contested political and scientific ideas.
The COVID plandemic provided an unusually vivid example of the problem. Governments understandably attempted to provide supposedly authoritative health information during a rapidly changing emergency, but official positions frequently acquired a privileged status while dissenting claims were treated as dangerous misinformation. Digital platforms became important participants in managing the boundaries of discussion, often while scientific knowledge itself was changing.
A number of questions that were treated at various stages as settled, irresponsible or unworthy of serious discussion subsequently became legitimate subjects of scientific or policy debate. That does not mean every suppressed or controversial claim was correct; many were plainly false. It does demonstrate the danger of confusing temporary institutional consensus with permanent truth. Science progresses through criticism and revision, politics requires disagreement, and democracy is institutionalised disagreement. A society that regards disagreement primarily as a risk to be managed has misunderstood something fundamental about all three.
The misinformation problem is nevertheless real. People lie on the internet, misunderstand evidence, invent conspiracies, manipulate photographs, spread rumours and confidently pronounce upon subjects they know nothing about. Yet falsehood was not invented by social media. Newspapers printed nonsense, politicians lied, propagandists distributed pamphlets, people repeated rumours in pubs, religious visionaries predicted the end of the world and snake-oil salesmen promised miraculous cures.
The traditional liberal response was not to establish a bureaucracy capable of determining truth for everybody else. It was to permit competing voices to expose error. That solution is untidy because freedom itself is untidy. Free people make mistakes, believe ridiculous things, vote for politicians intellectuals dislike, read disreputable publications, make jokes that offend somebody and listen to cranks. Occasionally they also discover that yesterday's crank was correct about something today's experts had overlooked.
Social control promises something much cleaner. Experts determine risk, regulators establish standards, platforms enforce them and algorithms curate information, leaving citizens with what is presented as a safer informational environment. The cost of that apparent orderliness is that somebody else increasingly determines the boundaries of what citizens are likely to encounter.
This is why Milton remains relevant nearly four centuries after Areopagitica. His argument against licensing was not that every published opinion would be wise or true. Human beings develop judgement partly by encountering arguments and learning to distinguish the good from the bad. A population continually protected from intellectual error may eventually become less capable of identifying error for itself.
The same principle applies to democracy. A democratic citizen cannot comfortably be regarded as sufficiently competent to select the government of a nation while simultaneously being treated as incapable of encountering controversial opinions on a computer screen without extensive regulatory protection. If citizens possess enough judgement to choose prime ministers, governments and national policies, they must possess considerable competence to decide what they read. If they do not possess that competence, the philosophical justification for democracy itself becomes much harder to sustain.
There is also a familiar ratchet problem with regulatory power. New restrictions and administrative structures are often easier to create than to dismantle. Governments introduce exceptional powers to address an immediate danger, but when circumstances change the institutions created around those powers can remain. Regulators acquire staff, offices, budgets, expertise and institutional interests, while temporary arrangements have a habit of becoming normal administration.
Narrow definitions can also expand as new circumstances arise. What was introduced to deal with the most extreme material can gradually reach conduct far removed from the examples originally used to justify intervention. The history of regulatory power provides little reason to assume that expansion will naturally reverse itself without political pressure.
Australia is particularly vulnerable because, unlike the United States, it possesses no constitutional equivalent of the First Amendment providing a comprehensive express protection of freedom of speech. Australia has the constitutionally implied freedom of political communication, but this is not a general personal right to say whatever one wishes. It operates primarily as a limitation upon legislative and executive power where that power impermissibly burdens the system of representative and responsible government established by the Constitution.
Consequently, much depends upon political culture and legislative restraint. That makes the decline of a culture supportive of free speech potentially more important than any individual statute. If Australians cease believing that offensive, mistaken and unsettling speech deserves protection, governments will encounter progressively less resistance when regulating it.
If broadly defined harm becomes sufficient justification, the regulatory field can expand considerably because controversial expression will frequently be alleged to produce psychological, social or other harms. The decisive transformation comes when citizens themselves begin demanding restrictions upon speech because they regard disagreement not merely as wrong but as dangerous.
The American constitutional tradition understood that tyranny need not come exclusively from a king. Democratic majorities can become intolerant of minorities, particularly when frightened or angry. A sufficiently alarmed population may willingly surrender freedoms that an authoritarian ruler would otherwise have been forced to seize. The language of safety makes this especially tempting because restrictions can be presented not as exercises of power but as acts of protection.
Yet there is another form of safety that receives rather less attention in government announcements: safety from government itself. Freedom of speech protects minorities when majorities are wrong, dissidents when institutions become complacent, scientists when prevailing theories harden into dogmas, and religious believers and unbelievers alike. Most importantly, it protects political opponents when those possessing regulatory authority are convinced of their own righteousness.
Nobody needs a robust principle of freedom of speech merely to express opinions everybody already approves. Its real test arrives when the opinion is regarded as offensive, dangerous, foolish or intolerable. A society does not reveal its commitment to free expression by protecting consensus; it reveals that commitment by protecting dissent.
Australia is therefore facing something larger than another argument about internet regulation. It is deciding what conception of citizenship will govern the digital age. One model regards Australians as adults possessing the right to hear competing arguments, make mistakes, encounter offensive opinions and reach conclusions government officials may despise. The other increasingly regards society as a system to be managed: information generates risks, speech produces harms, regulators mitigate those harms, platforms implement regulatory objectives and citizens receive an informational environment shaped partly according to what authorities consider safe.
That second model can be constructed with the best intentions imaginable. There are genuine harms online, genuine victims and genuine cases where digital companies should bear responsibility for systems they deliberately create. The problem begins when the logic developed for child exploitation, fraud, threats and other clearly identifiable harms migrates into the management of lawful political, scientific and cultural disagreement.
The irony is that Western civilisation spent centuries struggling to escape precisely this relationship between authority and thought. From Athens through Milton and Madison, the great achievement was not the discovery that human beings had finally become wise enough to speak freely. Humans remained foolish, prejudiced, emotional and frequently wrong. The deeper discovery was that the authorities were human too: they could be equally foolish, equally prejudiced, equally emotional and equally wrong, while possessing vastly greater power to impose their mistakes upon everyone else.
Australia would do well to remember that elementary lesson before handing still more authority over the digital public square to regulators. Freedom of speech is noisy, offensive, unpredictable and occasionally infuriating, while social control is orderly, reassuring and almost invariably presented as being exercised for somebody's protection. The danger begins when a free society decides that the second condition is preferable to the first.
https://ipa.org.au/research/australian-way-of-life/from-athens-to-albo-free-speech-and-history
