Can Australia Really Switch the Algorithm Off?

My inbox is full of the same question. Parents, principals, and wellbeing coordinators are all asking the same thing.

Is this real? Could our government actually make the platforms turn the algorithm off?

Here is what is on the table. Chanel Contos, who founded Teach Us Consent, used her National Press Club address to ask the federal government to make algorithmic feeds opt-in rather than opt-out. Her Fix Our Feeds campaign describes an experience that will feel familiar to anyone who was online a decade ago. Posts in the order they were published.  Content from the accounts you actually chose to follow. Explore pages built on a broad national average rather than a minute-by-minute reading of your own scrolling. And one clear switch: being profiled is something you say yes to, rather than something done to you while you are looking the other way.

The difference between opt-in and opt-out sounds like a technicality; however, it is the entire thing. If the algorithm runs by default and the off switch is buried three menus deep, almost nobody will find it and the harm continues exactly as before. Contos put it perfectly it is too easy for big tech to design schemes that look like structural reform but actually suit them just fine. She has taken the principle of affirmative consent she spent years carrying into schools and applied it to the feed. Consent has to be active, informed and easy to withdraw, whether we are talking about a bedroom or a recommender system.

Social Services Minister Tanya Plibersek has said that when a young boy in Australia opens a new social media account, it takes an average of 23 minutes before the algorithm begins feeding him misogynistic and violent content. Twenty-three minutes. The second national plan to end family and domestic violence is being drafted, and for the first time the algorithm sits inside its scope. The Greens have a Fix Our Feeds Bill before the Senate. Communications Minister Anika Wells has called recommender systems deliberately opaque. An exposure draft of the Digital Duty of Care is due within weeks. The Prime Minister wants Australians to have more control over what is served to them, although he has not gone all the way to opt-in. Europe already does a version of it under the Digital Services Act, every very large platform, Meta, TikTok, YouTube, X, must offer at least one version of each recommender system that is not based on profiling. That is why European users have a chronological feed sitting in their settings right now. The European Data Protection Board has gone further, saying the two options must be presented equally with no nudging toward profiling, and that while a user sits on the non-profiling feed the platform must stop gathering data to profile them.

So the machinery exists and the platforms built it, shipped it and maintain it. What Europe has not done is make the non-personalised feed the default, and that single step is what Contos is asking Australia to take. When someone tells you this cannot be done, what they usually mean is that it has not been made compulsory yet.

Australia is a reasonable place to try. We legislated the News Media Bargaining Code. We brought in the world first under-16 restriction. The eSafety Commissioner already sets Basic Online Safety Expectations. We have shown, more than once, that we will write law the platforms would rather we did not write.

Personalised recommendation is the business model, so expect a fight over revenue dressed as a fight for small creators, businesses and Australian publishers, who genuinely do rely on algorithmic distribution to be found. That concern is real, and the legislation should answer it rather than wave it away. Recommenders are also how platforms surface crisis resources, community notes and counter-speech, so a blunt off switch can degrade the good alongside the harm. The non-profiling feed has to keep carrying those protections. And the biggest risk is the compliant but worse feed. Slower, uglier, missing features, technically legal, quietly unusable. Europe has seen that pattern already, so feature parity and anti-nudging belong in the statute itself.

If I were designing it, I would build on five things. A statutory default of a chronological, follow-based feed. Personalisation enabled only through a standalone consent flow, never pre-ticked, never bundled into terms of service, revocable in one tap. Feature parity and no dark patterns. No data gathered about how a person scrolls while they are opted out. And audit powers, independent researcher access, penalties calculated on turnover.

For the schools and families I work with, default off would change the risk profile of a child's first weeks on a platform more than almost anything else we have tried. It would not stop harmful content existing and unfortunately it would not solve everything. But it would end the situation where a thirteen year old is inducted into misogyny in under half an hour, because there would be no algorithm running until he asked for one.

Getting the design right is the heavy lifting here, and the conversation that is happening right now, is exactly when our voices are worth the most. If you have ever thought about sending a short, plain email to your local member please do that while the window is open.