Australia's government has introduced the first national law designed to let social media users turn off the algorithmic feed, and the design decision that will determine whether the law matters is not in the headlines. The draft legislation, called My Feed, My Way and framed as a digital duty of care, would require platforms such as Meta, X, TikTok, and Google to notify users and offer a genuine choice over their feed: opt in to algorithmically personalized recommendations, or opt out and see only posts from accounts they follow. Associated Press reported that fines for non-compliance could reach A$109.2 million, with the eSafety Commissioner handling enforcement. The bill was introduced to Parliament in the week of September 8, 2026.
Everything about the law that is contested is contained in a question the draft answers only partially: what is the default? In every regulatory regime that has ever offered an algorithm choice, from cookie banners to the European Union's Digital Services Act, the observable result is that most users keep the default. A genuine choice with an engineered default is not much of a choice at all, and the platforms know this better than the legislators do.
The duty of care that carries the bill
The algorithm provision rides on a broader legal frame. The bill would impose on platforms a duty to protect users, and particularly children under 18, from harmful content, including material promoting eating disorders, misogynistic content, pornography, the glorification of crime, and dangerous stunts. The government's argument is that the feed is not merely a recommendation engine but a delivery system for whatever the engagement-maximizing algorithm surfaces, and that a duty of care over content is unenforceable without giving users a way out of the delivery system itself.
That framing matters legally, because it answers the obvious criticism that the law regulates speech rather than systems. The government's position, stated by Communications Minister Anika Wells, is that platforms can still move fast but will no longer be allowed to break things. Prime Minister Anthony Albanese put the same point more directly: the reform is not about giving government control, it is about giving people control.
The opposition and free speech advocates have taken the opposite reading, warning that the measures amount to censorship or government overreach, with The New Daily reporting ministers comparing the objections to the resistance to 1970s seatbelt laws. The comparison is telling about the government's expectations: it believes the safety framing will survive the political fight, and it is designing the law so the safety provisions and the algorithm choice stand or fall together.
The precedent no one wants to repeat
Australia is not first. The EU's Digital Services Act has required algorithm opt-outs since 2024, and its rollout is the cautionary tale hanging over the Australian design. European users gained the right to switch to non-personalized feeds, but the switch was buried in settings, the default remained personalized, and usage of the opt-out has been modest enough that the provision is widely regarded as a compliance exercise rather than a change in how feeds work. Australia's draft law is explicit about notification, requiring platforms to tell new and existing users about the choice, which reads as a direct response to the European experience: the right existed, and almost nobody knew.
Australia also comes to this with its own regulatory history with the platforms, most recently the world-first ban on under-16s using major social platforms that took effect in December 2025. The algorithm bill is best understood as the second act of that policy: having kept children off the platforms, the government now wants to change what the platforms show everyone else. The consistency is political as much as regulatory. Australia has positioned itself as the jurisdiction most willing to legislate the social platforms' core mechanics, and the algorithm feed is the core mechanic that remains.
Where the legal fights will actually happen
The bill's operative language will determine everything, and three drafting questions are already visible. First, the default. If platforms may set the personalized feed as the default and merely offer the opt-out, the law will replicate the European result. If the default must be the chronological feed, the platforms will argue the law has effectively banned personalization in a country of 26 million people, and the constitutional and trade arguments begin. Second, the definition of the choice itself. The draft contemplates users seeing posts from friends and creators they follow, but platforms algorithmically rank even chronological feeds, insert recommendations, and fill feeds with suggested accounts; every one of those mechanics is a potential loophole unless the drafting closes it. Third, the interaction with the duty of care. If platforms are liable for harmful content in personalized feeds but not in chronological ones, the economics of the default shift in the government's favor without the government ever having to mandate it.
The platforms' likely strategy is visible from the EU precedent: comply with the letter, bury the switch, and keep the default. The Australian government has anticipated that strategy, which is why the notification requirement is so prominent in the draft. Whether the drafting is strong enough to defeat it is what the committee process and the inevitable court challenges will test.
What the law would mean in practice
For an Australian user, the working difference is this: an opt-out feed shows the people you chose to follow, in the order they posted, and nothing else. For the platforms, the difference is existential, because the personalized feed is the entire advertising business. A user scrolling a chronological feed of followed accounts generates dramatically less engagement and dramatically less data than a user in the recommendation loop, and engagement is the product sold to advertisers. A law that meaningfully shifts users out of the loop does not regulate content; it regulates the business model, which is why the platforms will fight the defaults harder than the fines.
The A$109.2 million maximum penalty is large by Australian standards and small by platform standards, which is the final clue about where the real enforcement pressure is meant to come from: not the fine, but the design requirement. The bill's bet is that a visible, mandated choice, combined with a duty of care that makes the personalized feed itself a legal risk, will move user behavior in a way that ten times the fine could not.
The outcome that matters is not whether Australians can switch off the algorithm. The bill assures that. The outcome that matters is how many of them do, and that number will be decided by the default, which is still, as of this week, unwritten in the final text.
What enforcement without the fine actually looks like
The A$109.2 million headline penalty understates the law's enforcement design, because the bill gives the eSafety Commissioner something more useful than a fine: a standing power to investigate how the choice is presented. The regulator that administered the under-16 ban now gets the algorithm law, and the two statutes share a philosophy. Both regulate design rather than content, both put the compliance burden on the platform's architecture, and both treat the interface as the thing being regulated.
That is why the notification and choice requirements are drafted the way they are. A law that merely created a right to opt out would repeat the European result, where the right exists and usage is low because discovery is low. The Australian draft obliges platforms to notify new and existing users, which converts the opt-out from a buried setting into a presented decision. Whether a presented decision changes user behavior at scale is the empirical question the law is betting on, and it is the question the platforms' own product teams will answer fastest, because they already know what happens to engagement when users are actively asked which feed they want.
The history of the under-16 ban also matters to how this law will be complied with. Australian platforms did not fight the ban in court; they built age assurance systems and absorbed the cost, in part because the law's framing made defiance politically expensive. The algorithm law is designed for the same posture, which is why the duty of care is written so broadly and the opt-out sits inside it. A platform that wants to challenge the law in court has to explain why it should be allowed to keep children in a system that the parliament has now twice regulated, and why a visible feed choice is an unreasonable burden. That is a harder argument than any single provision of the bill suggests, and the platforms know it.
The bill's fate will be decided in a committee process the platforms will lobby hard, and the final text will show exactly where the lobbying landed: in the default, in the definition of the feed, in the carve-outs for search and music and messaging. Every carve-out is a door, and every door is a product that escapes the choice. The version of the law that matters is the one that survives contact with the platform industry, and that version does not exist yet. What exists is the clearest statement any democracy has made that the algorithmic feed is a regulated system, not a neutral one, and that the user deserves the switch. Whether the switch is wired to anything is the question the drafting still has to answer.
Primary sources
- Associated Press reporting on the opt-out announcement and the digital duty of care.
- The New Daily coverage of the censorship dispute and the minister's response.
- 1News reporting on the opt-out tools and the bill's introduction.
- bdnews24 for the world-first framing and penalty figures.