When policy gets ‘louder, not clearer’: what Australia's social media ban reveals about legislative design

Professor Giliberto Capano
ANZSOG Academic Fellow | University of Bologna

Australia’s 2025 social media ban for under-16s has not yet resulted in a single penalty against a platform. Yet the federal government is proposing to double the maximum fine for non-compliance from $49.5 million to $99 million and give the eSafety Commissioner stronger information-gathering powers.

The proposal has reignited debate about the effectiveness of a law aimed at changing the behaviour of powerful, multinational businesses. Some critics question the value of increasing penalties before enforcement action has established what constitutes non-compliance. Others point to continuing levels of under-age social media use as evidence that implementation remains challenging.

The discussion raises a broader question, though, that extends beyond social media regulation: why do some government policies struggle to be implemented even when their objectives seem clear?

Part of the answer lies in choices made at the drafting stage, not in how the law is later enforced. Through an ANZSOG research fellowship, I analysed the design of the Online Safety Amendment (Social Media Minimum Age) Act 2024 provision by provision. That analysis formed part of broader research into how legislative design influences implementation outcomes.

Policy debates often focus on outcomes after implementation. Yet many implementation problems can be traced back to design choices embedded in legislation long before a policy takes effect. Viewed through that lens, many of the concerns now being raised about the social media ban look less like separate problems and more like symptoms of underlying design choices.

Some 4.7 million under-16 accounts have been deactivated on platforms. However, a July report from the eSafety Commissioner found that 81.5 per cent of 10–15-year-olds were still using age-restricted platforms — only a slight decrease from 85.9 per cent prior to the ban. These figures are both accurate because they measure different things: the legal obligation concerns account ownership, whereas the stated purpose of the ban is to protect children from the harms associated with social media use. The law never specifies which of these counts as success.

The government has accused tech companies of deliberately undermining the law, while some parties argue larger fines miss the point entirely because major tech companies can simply absorb them as a cost of doing business.

Meanwhile, civil liberties advocates warn that the regulator’s expanded coercive powers could encourage over-blocking of legitimate content.

What is striking is that all these seemingly different criticisms point to the same underlying policy design problem: governments frequently create strong legal obligations in policy while leaving the content of compliance with these obligations undefined.

Australia’s social media ban illustrates this problem clearly. The reform’s central obligation requires platforms to take “reasonable steps” to prevent under-16s from holding accounts. Yet “reasonable steps” is not defined anywhere in the Act or in binding rules; it is only defined in guidance issued by the regulator ten months later, which platforms are not legally required to follow.

This explains the paradox of the missing fines: it is difficult to penalise a breach of a standard that has not yet been defined. It also explains why circumvention is so widespread: the definition of an adequate age check has not been specified either. It even explains the censorship alarm: when the law is vague, and the penalties are severe, platforms protect themselves by over-blocking. Vagueness plus pressure produces exactly the caution that critics fear.

To be fair, the expanded information-access powers address a genuine weakness: until now, the regulator has relied heavily on information provided by the platforms themselves. That is real progress. However, it sharpens the tools around an obligation that remains unclear.

The law also provides little guidance about how success should be judged. While it clearly identifies the target population (under-16s), it does not specify what outcome would demonstrate that the policy is working. How will policymakers determine whether the policy is achieving its intended objective and over what timeframe? The statutory review is designed to assess whether the enforcement framework is operating effectively, rather than whether the policy’s underlying objective has been achieved.

Whether the social media ban ultimately reduces harm remains an open empirical question. My concern here is not the goal, but whether the law’s design allows policymakers to determine if the goal has been achieved.

Furthermore, the bill ignores a capacity gap at the heart of the scheme. Two regulators are responsible for making this law work. The eSafety Commissioner has a dedicated team of more than forty people. In contrast, the Privacy Commissioner, who must ensure that the age-verification data of millions of Australians is used once and then destroyed, has far fewer dedicated staffing resources. The same office once spent seven years pursuing a single global platform over a single data breach. The bill strengthens the well-resourced regulator and gives nothing to the other: its own explanatory memorandum declares ‘no financial impact’.

The question is not whether either regulator is effective. It is whether the implementation burden created by the legislation is matched by the resources available to discharge it.

This is the pattern that my analysis predicts. When a government leaves a compliance standard open without a binding process to define it, it surrenders the ability to clarify the law. The only lever left is volume: higher penalties, stronger powers and tougher rhetoric.

Many implementation failures can be traced to design weaknesses that are identifiable before legislation is enacted. Australia’s social media ban demonstrates several of them, including an undefined compliance standard, limited success measures and a mismatch between responsibility and regulatory capacity.

The lesson in all of this is not that the policy’s objectives are unattainable. It is that ambition alone is not enough. Clear compliance standards, meaningful measures of success, and implementation capacity that match regulatory responsibilities are all design choices and can be addressed before legislation takes effect. A law that does not specify what success looks like can only ever get louder, not clearer.

Read more by Professor Giliberto Capano:

Designing robust and micro-calibrated policy: An ANZSOG Insights Paper examining the Job-Ready Graduates Package and Social Media Minimum Age Act to show how small design decisions can create major implementation challenges and shape long-term policy outcomes.

A diagnostic framework for robust and micro-calibrated policy design: A companion paper to the Insights Paper, presenting a practical framework for assessing whether policy design is likely to be robust, adaptable and capable of achieving its objectives under changing conditions.