Labor's online safety bill would hand ministers too much power

The elastic phrase at the centre of the bill

Labor's Online Safety Amendment (Digital Duty of Care) Bill 2026 rests on four words that carry an enormous amount of weight. Under section 26A(2)(a)(i) of the amended Online Safety Act 2021, platforms must identify "all reasonably foreseeable risks" of harm as part of mandatory risk assessments.^[1] Any practising lawyer asked to define the outer boundary of that obligation would find it difficult to locate one.

The phrase is borrowed from workplace health and safety law, where it functions because factory hazards are finite and physical. Applied to online content, it becomes an instruction to build a removal machine and keep feeding it new harm categories until legal exposure disappears. Lawful speech tends to be what gets caught in that process.

The child-safety case, and where it ends

The draft bill requires platforms to protect children from content promoting eating disorders, content promoting hostile ideas about women, pornography and content that glorifies crime.^[2] Those are concrete harms, and the case for regulating algorithmic amplification to minors is a serious one that few people dismiss outright.

The bill does not stop at child protection, though. The same framework that targets eating-disorder content also hands a minister ongoing power to write new harm categories into law without returning to parliament. That is a structurally different thing from a targeted child-safety measure, and the government's framing tends to blur the distinction.

Liberal senator Sarah Henderson told the CPAC conference on 11 October that Greens leader David Shoebridge had cited the views of Pauline Hanson and the Coalition and said: "we need digital duty of care so that those messages aren't amplified and they don't become the content that kids are digesting when they flip open their phone and wonder what's happening in the world." Henderson said Shoebridge has since tried to walk those comments back.^[3]

Ministerial discretion is the structural problem

Under sections 25C(2), 25D(2) and 25G(2) of the amended Online Safety Act, the minister may by legislative instrument determine additional categories of seriously harmful material, conduct harmful to children, and design features taken to have negative behavioural impacts.^[1] There is no statutory ceiling on those additions, and no requirement to return to the Senate for approval of each new category. Legislative instruments move faster and attract less scrutiny than primary legislation.

The Australian Human Rights Commission welcomed the duty of care concept but recommended that definitions of harm be made clearer and that ministerial rule-making powers be narrowed to protect freedom of expression, in a submission lodged on 23 September 2026.^[4] The Commission's concern goes directly to the architecture of the bill, not to its stated objectives.

Liberal senator Sarah Henderson said: "The Albanese government's digital duty of care raises the grave prospect of lawful voices being suppressed under the masquerade of online safety."^[3] That is a Coalition position, and readers can weigh it accordingly. The Australian Human Rights Commission's submission reaches the same destination from a different starting point.

Governments routinely write regulatory frameworks with ministerial discretion built in, then use that discretion in ways the original debate did not contemplate. With a bill this broad, the concern is less about immediate bad faith and more about what the power looks like in the Act across successive governments, with no clear limit on what qualifies as a harm worth addressing. The Human Rights Commission said the definitions need tightening. Consultation opened 8 September 2026 gave the government a chance to hear that case in full.