What the Digital Duty of Care Draft Gets Right, and the One Thing It Cannot Do

On Tuesday, the Government released its draft Digital Duty of Care, and with it a proposal that would let Australians switch off the algorithm. The Prime Minister has called it My Feed, My Way. I said publicly this week that algorithmic choice is the most consequential idea to enter Australian online safety policy in years, and I meant it.

Then I read the draft, and the feeling I have been trying to argue myself out of for 24 hours is that, as it currently stands, this is a tick in a box, a quick win.

The Digital Duty of Care and the under-16 minimum age were unveiled in the same fortnight in November 2024, as a two-part, world-leading package. Only one was a bill; the social media minimum age law got the legislation, the headlines, the international coverage, and a commencement date. The duty of care got a media release. The Digital Duty of Care was notably absent from the Bill. It read then, and it reads now, as the systemic-sounding companion to a popular measure. Enough to answer anyone who said the government was doing nothing structural, without yet having to do the structural thing.

Nearly two years later, it is still a draft.

There has been a statutory review, an issues paper, a public consultation survey, a framework paper in May, and now, this week, an exposure draft released for further consultation. Not a bill before the House, a draft, out for comment. Two years is a long time in policy, but it is a much longer time in a childhood. A ten-year-old who was in year five when this was announced is in high school now, and every one of those years was spent inside the systems the duty was supposed to govern.

A duty of care that names harms but cannot compel a company to hand over its own risk assessment is a statement of values, not a regulation. Nearly everything the world knows about how these companies weigh a child's wellbeing against a quarterly number came out through litigation or a whistleblower. It came out because a court could compel it, or because someone inside walked out with the files. No regulator in the world has been handed powers that strong. An Australian parent has one path: lodge a complaint with eSafety and wait. eSafety is a serious agency doing serious work, and it acts within the powers it is given. But what it is given is decided elsewhere, and it can only use the powers Parliament writes and the resources Parliament funds, and neither of those is in a parent's hands or in the Commissioner's.

I keep returning to the fact that the decision to call this a digital duty of care, rather than simply a duty of care, was itself a political choice. A duty of care already exists in many, many ways. It is one of the oldest ideas in the common law, and we apply it without ceremony to builders, bus drivers, and the people who manufacture prams, pool fences and children's cough syrup. Nobody had to invent a new category of law to make a manufacturer answerable for a product that hurts a child. The general duty was enough, because the general duty comes with a court attached.

Calling it digital does something specific. It lifts the technology sector out of the ordinary law of negligence and sets it down inside a bespoke regime, one drafted for the industry, consulted on with the industry, and enforced by an agency instead of a judge. It sounds like precision but it works like separation. And it explains the gap I described earlier, because once you are inside a bespoke regime, the parent is no longer a plaintiff. They are a complainant. The difference between those two words is the difference between a subpoena and a form.

The draft lets the Communications Minister add new categories of harm by regulation, without returning to parliament, where she is satisfied that serious harm to adults or harm to children is involved. The Senate can disallow, but concern about this has come from every direction, which is itself telling given that constitutional lawyers, civil liberties organisations, and the Coalition, whose shadow minister has called it an absolute threat to free speech.

I do not share that framing, and I want to put it the other way around, because almost nobody that I have read so far is. A scope that a minister can widen is a scope a minister can decline to widen. The harms facing our children are not static. If the list of what platforms owe a duty about depends on the attention and the appetite of whoever holds the portfolio in five years, then children’s safety has been made a matter of ministerial discretion rather than statutory right.

A single, systems-focused duty is exactly what the Online Safety Act review recommended, and exactly what those of us working in this field have been asking for. It moves regulation off the endless, exhausting business of taking down one post at a time and onto the machinery that decides which post reaches which child. The draft has real things in it. Penalties above one hundred million dollars. Foreseeable-harm risk assessment and algorithmic choice, with platforms required to prompt people about it more than once. The draft, as it stands, is doing double duty as a governance-sounding companion to a politically popular ban, still not law two years after it was announced, wearing a European label without the European machinery, and vesting the power to define its own scope in a single office.

Whether it becomes more than that depends entirely on what survives consultation.

There are five things I will be watching for, and I would encourage every school, every parent body and every clinician who deals with the aftermath of these systems to watch for the same five:

1. Independent regulator powers to compel, not request. 2. Mandatory independent risk assessments with published methodologies, so the working can be checked by someone other than the company that did it. 3. Researcher access to platform data, because we cannot regulate what only the platforms are permitted to measure. 4. A private right of action, so the duty means something to a family and not only to a department. 5. A harms list defined tightly enough in the statute that it cannot be quietly narrowed by a future minister who would rather not have the fight.

.