Australia is preparing to go further than any country in the world in regulating how social media platforms target children, with proposals for what amounts to an "off switch" for the algorithms that keep young users scrolling.
The plan would give regulators the power to require platforms to disable personalised recommendation systems for under-18s — the engagement-maximising software that decides what each user sees next.
It is the latest escalation in Canberra's running battle with the technology giants, following its world-leading ban on social media accounts for under-16s.
Why algorithms are the target
Campaigners and researchers have long argued that the real engine of social media harm is not the apps themselves but the recommendation algorithms beneath them, which learn what each user cannot resist and serve more of it.
For adults, that might mean an endless feed of outrage or trivia. For children, the evidence increasingly suggests, it can mean being drawn into content about self-harm, eating disorders, extreme body image and worse — served up not because they searched for it, but because the algorithm calculated it would hold their attention.
The Australian proposal would not ban the platforms for teenagers. Instead, it would strip out the personalisation, leaving young users with chronological feeds of accounts they actually follow.
The industry's response
The technology companies are expected to fight the measures hard. Recommendation algorithms are the commercial heart of their businesses, driving the engagement on which advertising revenue depends.
Industry bodies have argued that such rules are technically unworkable, easily evaded and an overreach by government into product design. They point to existing safety features and age-assurance tools as evidence that the sector can regulate itself.
Critics of the industry note that similar arguments were made against the under-16s ban — which Australia passed anyway, and which other countries are now studying closely.
A global test case
What happens in Australia will be watched around the world. The UK has its own Online Safety Act, and Ofcom is already pressing platforms over algorithmic amplification of harmful content to children.
In Europe, regulators are using the Digital Services Act to investigate recommendation systems, while in the United States a patchwork of state laws is emerging in the absence of federal action.
If Canberra can make an algorithm off switch work — technically and legally — it will provide a template. If the industry defeats it, that too will be noted.
The families behind the policy
Driving the proposals are bereaved parents who have become formidable campaigners, telling parliamentary inquiries how their children were served harmful content by systems designed to maximise their attention.
Their argument is simple: these are the most sophisticated persuasion machines ever built, and children are being left alone with them.
The platforms say they are investing heavily in safety. The Australian government's reply, increasingly, is that investment is not the same as change — and that where persuasion fails, regulation will follow.
The technical details will decide whether the policy succeeds. Chronological feeds are simple in principle, but platforms could respond with design changes that preserve engagement in other ways, and regulators will need the expertise to police what happens inside proprietary systems. Australia has signalled it is prepared to fund that capability, building a dedicated technology unit within its eSafety regulator. Legal challenges are considered certain, with companies expected to argue that the rules exceed the regulator's authority and burden legitimate speech. The government points to its record: it has fought the platforms before, over news content and age restrictions, and has largely prevailed. Ministers say they expect the same outcome this time — and that other governments are already asking how the model could be copied.
