The Albanese government's proposed Digital Duty of Care legislation represents a potentially profound threat to freedom of speech in Australia. Its danger does not lie principally in an explicit prohibition upon conservative political opinions, or even in the prospect of government officials ordering the removal of particular articles. The more serious danger is that it could establish a system in which private technology companies are encouraged to censor lawful political speech themselves, under the pressure of vaguely defined legal obligations and enormous financial penalties. It is a sophisticated form of indirect censorship, one that can operate without the government ever admitting that censorship is its objective.
The central mechanism is deceptively simple. Online platforms would be required to take reasonable steps to prevent specified categories of serious harm. Such a duty sounds unobjectionable when presented as protecting children from exploitation, pornography, bullying or other genuinely dangerous material. But the political significance lies in how broadly harm can be defined, how those definitions can develop through regulation, and how platforms are expected to respond when the consequences of getting the decision wrong are potentially ruinous.
The proposed framework has attracted criticism because of its open-ended concepts of harm and the possibility of additional categories being designated through ministerial powers. The precise reach of those powers depends upon the final statutory wording and associated instruments. Nevertheless, the underlying danger is clear. When the boundaries of harmful conduct are uncertain, and regulators possess substantial discretion in interpreting them, private companies must attempt to anticipate what future enforcement decisions might require.
This becomes especially troubling when combined with civil penalties potentially reaching approximately $100 million for serious contraventions, depending upon the applicable penalty provisions. A large technology company does not need to be ideologically committed to censorship to become an enthusiastic censor. It needs only a legal department capable of calculating financial risk.
Suppose a controversial political article criticises government immigration policy, challenges prevailing views about gender, questions official public-health recommendations or argues that multicultural policies have failed. Such arguments may be lawful contributions to public debate. Yet they may also generate complaints alleging hostility, psychological harm, discrimination or misinformation. Where the regulatory boundaries are uncertain, a platform must decide whether defending the publication is worth the possible expense and regulatory scrutiny.
The commercially rational answer may be to remove it, restrict its circulation, attach warnings or make it inaccessible to certain users. Why risk a major financial penalty for an article that generates little revenue? Why defend a small independent publisher when removing the offending material is cheaper and easier?
This is how censorship can develop without an official censor issuing direct commands. Government establishes the regulatory framework, private companies interpret their exposure, and automated moderation systems implement the resulting restrictions. The state need not formally prohibit a political opinion if it can create incentives that make private intermediaries reluctant to distribute it.
The arrangement also provides political deniability. When an article disappears, a government minister can insist that no official ordered its removal. The platform can claim that it was merely applying its safety policies. The regulator can maintain that it has not prohibited the underlying opinion. Yet the citizen has still lost access to lawful political expression, and the author may have little effective opportunity to challenge the decision.
There is a fundamental asymmetry in such a system. A platform that fails to comply with its legal obligations may face substantial penalties. But a platform that unnecessarily suppresses lawful political speech may face no equivalent financial consequence. The ordinary citizen whose article is removed has neither the resources nor the legal leverage of a multinational technology corporation. Even where internal appeals exist, they can be slow, opaque and ineffective.
The predictable result is over-compliance. Platforms have an incentive to remove too much rather than too little. This tendency is especially likely where the rules are ambiguous, penalties are severe and the commercial benefits of defending controversial expression are limited. The system can gradually narrow the boundaries of acceptable discussion without requiring any formal change to the criminal law.
The political implications are considerable. Almost every important public controversy involves allegations of harm. Immigration debates generate claims of racial hostility and social division. Arguments about transgender policies produce allegations of psychological injury and discrimination. Religious criticism can be described as hateful or offensive. Questions about public-health policy may be characterised as dangerous misinformation. Criticism of government institutions can be presented as undermining public trust.
None of this means that genuinely unlawful threats, harassment or incitement should be protected. The distinction between legitimate political argument and unlawful conduct remains essential. But a free society must not allow subjective claims of offence or distress to become substitutes for clearly defined legal standards. Otherwise, the more controversial an opinion becomes, the greater the incentive to suppress it.
The problem is intensified when ministers or regulators can influence the categories of material and conduct subject to the system. The government of the day may possess entirely sincere intentions concerning child protection. But legislation must be judged not merely by the intentions of its present supporters. It must also be judged by the powers it creates, the incentives it establishes and the ways future governments might employ those powers.
Once an administrative system exists for influencing what online platforms classify as harmful, its potential applications need not remain confined to the original political justification. Future governments may have different priorities, different understandings of social harm and different attitudes towards dissent. The machinery will already exist. Only the interpretation and enforcement priorities may need to change.
Australia's constitutional arrangements make this particularly concerning. Unlike the United States, Australia has no general constitutional guarantee of freedom of speech equivalent to the First Amendment. The High Court has recognised an implied freedom of political communication, but this is principally a restriction upon legislative and executive power rather than a comprehensive individual right to speak freely. Laws burdening political communication may survive constitutional scrutiny where they satisfy the applicable legal tests.
Australians therefore cannot assume that courts will automatically invalidate every law that indirectly discourages political expression. Nor does the implied freedom necessarily provide an effective remedy against every moderation decision made by a private company. A regulatory framework that encourages private censorship may consequently produce restrictions that are difficult for ordinary citizens to challenge.
There is a bitter irony in the fact that Google has warned about the implications of the proposed legislation. Google is hardly a disinterested champion of privacy, given the enormous quantity of personal information collected across its services. Yet the company has raised concerns that broadly framed online safety obligations could extend into private communications and file storage, potentially encouraging intrusive monitoring of material that Australians reasonably expect to remain confidential.
When even Google warns that government regulation may require excessive surveillance, Parliament should pay attention. The fact that a corporation has commercial interests does not automatically invalidate its technical assessment of the consequences of legislation. Indeed, the companies expected to implement these obligations may understand better than politicians how difficult it is to distinguish supposedly harmful content from lawful communication across billions of messages, documents and online interactions.
The danger is not confined to surveillance of private correspondence. It extends to the architecture of the internet itself. Increasingly complex compliance obligations favour large corporations that possess substantial legal departments, sophisticated automated moderation systems and the financial resources to absorb regulatory costs. Smaller platforms and independent publishers may struggle to satisfy the same requirements.
The result could be greater concentration of control over public communication. Governments may claim to be restraining the power of major technology companies while simultaneously creating regulatory conditions that strengthen their dominance. Independent voices become increasingly dependent upon the decisions of a small number of powerful intermediaries.
Australia has already witnessed intense controversy over government attempts to regulate online misinformation. The Albanese government's earlier misinformation legislation was withdrawn in 2024 following substantial political opposition and concerns about freedom of expression. Critics now fear that aspects of the same regulatory philosophy could reappear through a broader duty-of-care framework, where the focus shifts from prohibiting particular statements to requiring platforms to manage categories of harm.
These are not legally identical approaches, and the proposed duty should not be misrepresented as an automatic ban upon political disagreement. But the concern is that both approaches may place considerable discretion over lawful expression in the hands of regulators and private platforms, while making the boundaries of acceptable speech increasingly uncertain.
The wider age-verification experiment provides a disturbing illustration of what happens when access to information becomes dependent upon automated compliance systems. Australians attempting to read lawful online publications can encounter demands to prove that they are adults, even when there is no serious question about their actual age.
In one recent case, an Australian pensioner repeatedly attempted to satisfy Substack's age-verification procedures, only to encounter failures and procedural obstacles over several days. A facial age-estimation system could not establish that the individual was over eighteen. The result was that a mature adult was effectively prevented from accessing material intended for adults because a computer system could not recognise the obvious.
That experience does not establish that the proposed Digital Duty of Care legislation caused the particular Substack failure. But it demonstrates the practical consequences of making access to lawful information conditional upon automated verification. A technology that performs adequately in statistical tests can still exclude real individuals, and those individuals may have little practical recourse when the system fails.
There is something profoundly wrong with a society in which an elderly citizen must convince an algorithm that he is legally entitled to read a publication. The presumption of liberty has been reversed. Instead of being free to access lawful material unless a specific legal restriction applies, the citizen must satisfy a technological gatekeeper before exercising an ordinary freedom.
The same logic threatens to spread through online communication more generally. The more activities that require age checks, identity verification, automated classification or compliance approval, the more the internet becomes a regulated permission system rather than an open medium of communication.
The government's defenders will argue that these measures are necessary to protect children and vulnerable people. There are genuine online dangers, and governments have legitimate responsibilities in responding to them. But the existence of a serious problem does not justify every proposed solution. Effective protection requires clearly defined obligations, proportionate enforcement, privacy safeguards and meaningful rights of appeal. It does not require the creation of a regulatory environment in which lawful political speech becomes an avoidable commercial liability.
The fundamental political question is whether Australians are prepared to accept a system in which government influences the boundaries of public debate indirectly through financial pressure upon private intermediaries. Such a system may be less visible than traditional censorship, but that does not make its effects less serious.
Indeed, indirect censorship may prove more difficult to resist precisely because responsibility is so widely dispersed. The government blames the platform, the platform blames compliance requirements, and the regulator points to general statutory obligations. The ordinary citizen is left confronting an opaque bureaucracy in which no single institution accepts responsibility for suppressing lawful expression.
A society does not require an official Ministry of Truth to develop a culture of censorship. It requires only vague regulatory standards, severe financial penalties, powerful intermediaries and insufficient protection for speech that those intermediaries find inconvenient or risky.
Australia should be moving in the opposite direction. Any online safety framework should be tightly confined to clearly identified harms, should explicitly protect lawful political communication and private correspondence, and should provide effective remedies for wrongful content removal. Ministerial powers to expand regulated categories should be narrowly limited and subject to meaningful parliamentary scrutiny.
Above all, the government must recognise that freedom of speech is not merely another administrative consideration to be balanced away whenever regulators identify a possible risk. It is a foundational condition of democratic accountability. Without the ability to criticise government policy, challenge prevailing orthodoxies and express unpopular opinions, citizens cannot meaningfully exercise political judgment.
The greatest danger of Australia's proposed online safety regime is therefore not that every dissenting website will disappear overnight. It is that the range of permissible public argument may gradually contract as platforms learn that silence is safer than controversy. Writers will moderate their criticisms, publishers will avoid difficult subjects, and citizens may never know what information has been withheld from them.
The censorship will not necessarily arrive with police raids, banned books or official lists of forbidden opinions. It may arrive through compliance departments, automated moderation systems, unexplained account restrictions and the quiet disappearance of material that somebody, somewhere, decided was too risky to publish.
That is censorship without visible censors. And for a country already lacking strong constitutional protection for freedom of speech, it is a danger Australians should resist before the machinery becomes an accepted part of everyday life.
https://www.macrobusiness.com.au/2026/11/the-end-of-australian-free-speech/