PHOTO: Pixelkult on Pixabay
On September 22, 2026, the US Embassy in Canberra submitted a formal objection to an Australian draft law. The embassy released a media statement submitting that measures forcing platforms to filter protected speech would likely be seen in Washington as “facilitating censorship”.
The target was the digital duty of care. The draft would require anyone responsible for an online service to ensure, “so far as is reasonably practicable”, a safe online environment. Striking the balance between censorship and fair regulation has proven itself to be a Sisyphean task for states across the globe.
Platforms headquartered in one country shape what users see in another, so any state that regulates them is, in theory, claiming authority beyond its borders. A duty of care can protect a teenager’s feed. The duty of care obligations can also reach information flows a government cannot otherwise touch. Governments write it into law for different ends, and the contest between them concerns who governs infrastructure that no single state controls.
Australia sits in the middle of the dispute, harnessing more influence than middle powers typically do: it was the first country to bar under-16s from major social media platforms, and its prime minister co-chairs a coalition of 23 states on children’s rights and AI. Its duty of care bill will show whether a middle-power democracy can set terms for companies that operate beyond its borders.
This article compares four jurisdictions: the EU, the UK, the US and Australia.
PHOTO: Two women relax by an urban fountain using smartphones, Surprising_Media on Pixabay
Who Is Responsible for Online Harm?
Duty of care is an old idea in law, and famously, it began with a ginger beer. Mrs May Donoghue found a decomposed snail in a bottle a friend had bought her, and she sued the manufacturer. She had no contract with him. In Donoghue v Stevenson in 1932, the House of Lords, then Britain’s highest court, ruled that a manufacturer must take reasonable care not to injure the people who end up using its products. The case is widely treated as the start of modern negligence law.
The digital version would shift the onus onto platforms to anticipate harm before it reaches users. Australia’s current system is reactive, it relies on complaints and takedowns.
Two harms drive the legislation. The first is children online. Australia barred under-16s from major social media platforms on December 10, 2025, and platforms had restricted about 4.7 million accounts within weeks. eSafety Commissioner Julie Inman Grant said some underage accounts remained active and it was too early to call compliance full.
The second harm is polarising content. Teach Us Consent founder Chanel Contos launched the #FixOurFeeds campaign in December 2025. Contos said young men and boys are funnelled into misogynistic online spaces, and that research shows extreme content reaching them within minutes of signing up. She spoke at the National Press Club, urging the government to make recommendation algorithms opt-in.
Last week, litigation started moving in Australia. MinterEllison filed a class action in the Federal Court on behalf of eight plaintiffs who say Facebook and Instagram exposed them to harmful content and kept them engaged through addictive algorithms while they were under 18. It could be the first Australian case to argue that a social media company harmed its users, and Meta — represented by Herbert Smith Freehills Kramer — is defending the claim. The Greens’ proposed right to sue under the bill would give plaintiffs like these a second, consolidated route to court.
In other jurisdictions, the harms have already been tested in court. In March 2026 a Los Angeles jury found Meta and YouTube negligent in a case brought by a woman who said using social media as a child caused her depression and anxiety. A day earlier a New Mexico jury ordered Meta to pay US$375 million (about AU$570 million) for failing to protect young users from child predators. A Meta Spokesperson says teen mental health is “profoundly complex” and cannot be linked to one app.
The UN sets standards and states set rules
While global standards have been established, enforcement remains fragmented. The UN Committee on the Rights of the Child adopted General Comment 25 in 2021, setting out how children’s rights apply online and what duties states and companies carry. But, this document does not grant any party the power to fine a platform.
Now, states are rushing to address that gap. Prime Minister Anthony Albanese told the UN General Assembly in September that more than 20 countries have brought in or are bringing in age limits on social media.
The European Union
Brussels moved first on enforcement. In December 2025 the European Commission fined X €120 million (about AU$210 million), the first penalty under the Digital Services Act. The Commission said X’s paid blue tick misled users and its advertising database was faulty. It also said X blocked researchers from its internal data. Fines can reach six per cent of a company’s annual worldwide turnover. X and its owner, Elon Musk, have challenged the decision in the EU’s General Court.
PHOTO: Dima Solomin on Unsplash
The Commission has since turned to how platforms are designed. In February 2026 it found preliminarily that TikTok’s addictive design breaches the Act, pointing to features such as infinite scroll. In July it also preliminarily found that TikTok’s default settings leave minors exposed to unwanted contact. TikTok can still respond before any decision is final.
Washington answered with visas. Less than three weeks after the fine, on December 23, US Secretary of State Marco Rubio banned five Europeans. They included former European commissioner Thierry Breton, who helped draft the Digital Services Act. Rubio accused them of leading “organized efforts to coerce American platforms to censor, demonetize, and suppress American viewpoints”. French President Emmanuel Macron called the bans “intimidation and coercion aimed at undermining European digital sovereignty”. The word sovereignty had entered the dispute.
The United Kingdom
The UK’s Online Safety Act 2023 shows how hard it is to enforce a law against a company based in another country. Ofcom, the UK communications regulator, fined the US message board 4chan £520,000 (AU$1 million) in March 2026 for failing to protect children from pornography and to assess the risk of illegal content. 4chan mocked the fine and has refused to pay earlier ones as enforcing the fine abroad would require the involvement of a US court.
The company, 4chan, also took the fight to American courts. 4chan and the forum Kiwi Farms sued Ofcom in Washington DC, arguing its demands interfered with their constitutional rights. In September the court dismissed the case, saying it had no jurisdiction and that the platforms could raise their defence in British courts. Ofcom said the ruling confirmed it is within its rights to investigate breaches of UK online safety law.
The United States
The US courts are testing the same question. Congress has debated the Kids Online Safety Act since 2022. The bill aims to legislate that social media companies should carry a legal responsibility to protect young users from harm. The Senate passed the bill 91 to 3 in 2024, and it still never became law.
On June 29, 2026, the House passed its own child safety package, the Kids Internet and Digital Safety (KIDS) Act, by 267 votes to 117. The bill combines 14 earlier proposals and adds rules for social media platforms and AI chatbots, the International Association of Privacy Professionals reports. It left out the duty of care, the clause that makes platforms legally responsible for protecting young users.
Authors of the Kids Online Safety Act, Senators Richard Blumenthal, a Democrat, and Marsha Blackburn, a Republican, have worked on their own online safety legislation. They called the House bill “a pale imitation” of Big Tech accountability. On August 5 the Senate Commerce Committee advanced a version that keeps the duty, noting it still needs approval from the full Senate. The Electronic Frontier Foundation, a digital rights group, opposes the bill and argues its age checks would weaken privacy for everyone.
Washington’s objection to Australia’s bill exists uneasily alongside its own debate over platform accountability. The dispute reflects an ongoing, broader divide over how far governments should go to protect users without restricting speech.
Platforms test every border
Platforms are not waiting for laws to pass. Meta, TikTok and X have taken Ofcom to a UK court over its demands for internal data. X’s witness statement called one request “the most burdensome information request” the company had received.
In Australia, X has made challenging jurisdiction a habit. In 2024 the Federal Court declined to extend an injunction that would have forced X to remove links to footage posted of the 2024 Wakeley Church stabbing for all users. The court noted that a US court would be highly unlikely to enforce such an order. Musk said the order would let one country control the entire internet.
Despite this, X has not won every case. In 2025 the Full Federal Court rejected X’s appeal and held that it had to answer a transparency notice about child abuse material. In May 2026 X was ordered to pay a $650,000 penalty for failing to answer that notice in full.
Platforms also have an ally with a diplomatic corps. The US Embassy’s submission warned that Australia’s rules could limit how American platforms operate and affect other markets if copied, and said they risk “disproportionate compliance burdens on American companies”. The embassy’s full submission is public.
Industry group DIGI, which represents Meta, TikTok and Snapchat in Australia, generally welcomed the idea of a well-crafted duty of care in principle in September. The argument has since moved from whether to how far.
Australia’s duty of care
Released in September, the exposure draft would replace Australia’s existing industry codes and safety expectations with a single legal duty, requiring anyone responsible for an online service to ensure, “so far as is reasonably practicable”, a safe online environment.
The eSafety Commissioner would enforce it. A company that breaches the duty could face a penalty of up to $109.2 million. Platforms could also be required to name a contact person in Australia, so the regulator has someone to reach. On October 15 2026, Communications Minister Anika Wells introduced the bill to parliament.
The bill moves regulation upstream. A regulator could ask whether a platform built a feature that could harm young users before a breach or complaint occurs. Platforms would have to protect users under 18 from content that promotes crime or eating disorders.
Contos launched the #FixOurFeeds campaign in December 2025 to make algorithms opt-in. The bill would require platforms to offer users a choice over their default feed: a personalised feed or one showing only accounts they follow, in the order posted. Contos asked for the algorithm to be off by default. The September draft stopped short of that, and the minister’s October 15 release says users will decide whether to opt-in.
Crucially, the eSafety Commissioner can only audit what platforms elect to disclose. To close that gap, the bill would give approved researchers, such as university academics, access to platform data. Enforcement only rests on the eSafety Commissioner’s office, which Wells said carries a heavy burden. A “private right of action” would distribute that burden more broadly, including through class actions brought on behalf of affected groups. Greens communications spokeswoman Sarah Hanson-Young said that without this avenue for redress, platforms have little incentive to comply. The government is still considering the proposal.
The limits of power
Critics focused on how much power the draft handed to ministers. Liberal communications spokeswoman Sarah Henderson called the plan “dangerously vague”. Labor then dropped the minister’s proposed power to define new and “foreseeable” harms after concerns it could lead to political censorship. Parliament would now decide on new harms, which Wells said would be slower. The bill also exempts professional news material and entertainment media. In light of the changes, Albanese said the draft is “not about censorship”.
PHOTO: A young person using a smartphone, Richard Williams on Unsplash
Reach is the harder question. As mentioned earlier, in the 2024 case against X, the Federal Court held that the eSafety Commissioner cannot require a foreign-hosted platform to remove content worldwide. The draft would let the regulator require a contact person in Australia, which would help serve notices. Despite the contact person requirement, the draft legislation does not create assets that could be sought.
Can Australia Make Its Rules Stick?
The bill was introduced to parliament on October 15, and the government wants it passed before the end of 2026. The Senate referred it the same day to its Environment and Communications Legislation Committee, which holds public hearings in early November and is due to report on November 20. If the bill passes, the rules would take effect in late 2027.
Three tests follow for the government:
The first is enforcement. A penalty of up to $109.2 million means little against a multinational with few assets in Australia.
The second is safeguards. Passing the bill means negotiating with the Greens, whose proposed right to sue platforms would add enshrined legal exposure to a regime the US Embassy already says burdens American companies. Every concession to the Senate crossbench on this will sow dissent on this front.
The third is collective action. Australia co-chairs a coalition on children’s rights and AI that now has 23 member states and eight UN bodies. Albanese shares the chair with Macron and Spanish Prime Minister Pedro Sánchez. Kenyan President William Ruto is the fourth co-chair. Indonesia joined after the General Assembly and will attend a follow-up summit in November. Membership does not bind any government to pass a law like Australia’s. Each member that legislates gives Australia a yardstick to assess against.
The remedies carry their own risks. Age checks require sensitive data, and in 2025 about 70,000 Discord users may have had government ID photos exposed after a breach at a vendor that handled age-related appeals. Australia’s privacy regulator, the Office of the Australian Information Commissioner, has issued guidance on destroying age-verification data once a decision is made.
Advocacy groups are calling for oversight to also limit officials. The Human Rights Law Alliance, a legal advocacy group, argues that the eSafety Commissioner should have to seek binding orders from the Administrative Review Tribunal or the Federal Court before imposing directions with large financial consequences. The Human Rights Law Centre wants an overarching human rights safeguard written into the bill. In this event, every safeguard slows enforcement, and slower enforcement leaves children exposed for longer. The government is yet to formalise the balance between these interests.
So, how much delay does a democracy accept to keep both firms and officials in check?
Each government says it is protecting users, and each regulator needs something from a company based abroad, such as data or a payment. X is contesting its EU fine. eSafety took three years in court to win a $650,000 penalty from X. So while the duty of care gives every regulator a new reason to ask, it gives none of them a new way to make a platform answer.
Montana Basile
Montana Basile is a Bachelor of Laws (Honours)/Arts (International Relations) student at Monash University. Her work sits at the intersection of law, public policy, and global security, with an interest in defence and strategic affairs. Alongside her academic pursuits, she has experience in the private legal sector. Montana is passionate about platforming women and LGBTQIA+ voices within legal and political spaces, and is involved in initiatives promoting inclusive leadership. She is engaged in youth advocacy with Youth UN Australia and the Folio Collective, where she promotes civic education for young people and strengthening public perceptions of civic participation.