Government bends ‘evidence’ to create Social Media laws
Australia's government went to the High Court to defend its teen social media ban — and may have handed its opponents their best argument.
When a government argues that scientific uncertainty should not delay action, it is making a legitimate claim. The precautionary principle has genuine intellectual standing. But when that same government then argues its proposed alternative policy would achieve the same result as the ban it rushed through, it has inadvertently made the opposition's case for them.
The government's own defence undermines the case for a ban
The concession came in the government's own court filing. Defending the ban against a challenge brought by Reddit, the government acknowledged that the "agreed facts" in the case establish no established causal link between social media use and the harms the ban is designed to prevent. What exists instead, in the government's framing, are "significant credible risks." Addictive behaviours, disrupted sleep, worsening anxiety and depression, suicidal ideation: these are real concerns, grounded in real evidence, and it would be wrong to dismiss them as invented. The question is whether a blunt instrument like a ban is what the evidence actually calls for, and the government's own legal filings suggest the answer is complicated.
Here is the problem the government has created for itself in court. Its defence argues simultaneously that the ban was justified by the severity of potential harm, and that its forthcoming digital duty of care legislation, which would require platforms to default away from algorithmic feeds and allow users to opt out of recommendation systems, would achieve "the same action." If both things are true, then the government had a less restrictive tool available that would have produced the same protective outcome for children. That is precisely the argument Reddit is making. A law that restricts freedom is much harder to defend constitutionally when a less restrictive alternative exists that the government itself concedes would work just as well.
A law that restricts freedom is much harder to defend constitutionally when a less restrictive alternative exists that the government itself concedes would work just as well.
The algorithmic opt-out framing is worth dwelling on. The government's proposed digital duty of care framework, covered in detail in earlier Bearing analysis of the Fix Our Feeds bill, would require platforms to offer chronological, non-algorithmic feeds by default. A separate piece examined where the actual power in that framework sits: not with users choosing their experience, but with regulators deciding which features count as harmful in the first place. Neither of those frameworks is without problems. But they are both more surgically targeted at the specific mechanism of harm — algorithmic amplification and engagement-maximisation design — than a blanket account ban.
Enforcement tells a different story than the headline numbers suggest
The government points to nearly five million accounts removed in December, including over 43,000 Reddit accounts suspended between December and March, as evidence the ban is producing "substantial practical action." That number is real. What is harder to establish is whether the action is producing the protective effect it was designed to produce. Children who want to access social media have continued to do so. The ban's own enforcement architecture depends on platforms self-policing, and the government has had to double the available fines to $99 million to try to force more diligent compliance. That is not a sign of a policy running smoothly on its own design logic.
The precautionary principle permits action — it does not choose the action
This is where the precautionary principle argument, though legitimate in principle, becomes slippery in practice. "Uncertainty about causation should not delay action" is a defensible position. But it is also a position that can justify almost anything, because uncertainty about causation is the default condition of complex social policy. The precautionary principle does not tell you which action to take, only that some action is permissible. Choosing the most restrictive available action, on the basis of uncertain evidence, when a less restrictive option is available, requires more justification than the principle itself provides.
What it does provide is political cover. A ban is visible. A ban can be announced, defended, and pointed to. An algorithmic redesign requirement is harder to explain to a press conference and harder to photograph. When the government introduced the ban it described it as "world-leading." That framing tells you something about the logic at work. Genuine harm-reduction policy does not usually reach for the word "world-leading" first.
The High Court will decide the constitutional question. But the evidential question is already answered: the government has conceded, in its own defence, that the scientific foundation for the ban's specific design is not settled. That concession does not make the ban wrong. It makes the choice to design it this way, rather than another way, a political decision dressed as a public health necessity.
Sources
Frequently Asked Questions
What did the Australian government concede in the High Court social media case?
The government acknowledged in its own court filing that there is no established scientific consensus linking social media use to mental health harm in teenagers. It argued instead that 'significant credible risks' — including disrupted sleep, anxiety, depression, and suicidal ideation — were sufficient to justify the ban under a precautionary principle.
Why does Australia's social media ban face a constitutional challenge?
Reddit is challenging the ban on the grounds that a less restrictive alternative was available — specifically, the government's own proposed digital duty of care legislation, which it conceded would produce 'the same action.' A law restricting freedom is constitutionally harder to defend when the government itself acknowledges a less intrusive option would achieve the same outcome.
What is the precautionary principle and does it justify the social media ban?
The precautionary principle holds that scientific uncertainty about causation should not delay protective action. It is a legitimate policy position, but it does not specify which action to take — only that some action is permissible. When a less restrictive option exists that the government concedes would work equally well, the principle alone cannot justify choosing the most intrusive available measure.
Is Australia's under-16 social media ban actually working?
The government cites nearly five million accounts removed since December as evidence of 'substantial practical action,' but account removal and protective outcome are different things. Children who want social media access have continued to find it, and the government has had to double maximum platform fines to $99 million to force greater compliance — not the profile of a policy running on its own design logic.
What would algorithm opt-out laws do differently from a social media ban?
Algorithm opt-out laws would require platforms to offer chronological, non-algorithmic feeds by default, targeting the specific design feature — engagement-maximising recommendation systems — most associated with harm. Unlike a blanket account ban, they would not prevent access to platforms entirely, but would instead change the default conditions under which content is served to users.