The government released an exposure draft of its digital duty of care laws on Tuesday, promising Australians over 16 the ability to switch off personalised feeds on social media.
The requirement is not in the draft. Under section 26(4), the Communications Minister "may, by legislative instrument, require specified online services to provide specified user empowerment tools", a term the bill defines to include "control over the kind of content that is recommended for the user". The words opt out, algorithm and notification appear nowhere as an operative obligation, and the draft does not require a chronological feed.
What the Online Safety Amendment (Digital Duty of Care) Bill 2026 does directly is impose a duty: a person responsible for an online service "must ensure, so far as is reasonably practicable, a safe online environment". Services must manage design features, run written risk assessments of foreseeable harms, and take measures to address them. Enforcement sits with the eSafety Commissioner, who can issue a warning, then a remedial direction, then pursue a civil penalty of 60,000 penalty units, which the government puts at $109.2 million.
Prime Minister Anthony Albanese and Communications Minister Anika Wells announced it at Parliament House under the title "My Feed, My Way". Platforms would have to notify new and existing users and let them choose whether their default feed carries recommended content or only the accounts they follow.
“This is not about giving government control, it is about giving people control," Albanese said. Asked at the press conference whether the laws would let regulators police misinformation, he said: "Let's be very clear about what this is about. It's not about censorship." He said more than five million accounts had been removed or deactivated under the under-16 ban, and that platforms breaking the new rules "will face significant penalties of more than $100 million".”
Wells put it in consumer terms. "Just as we have basic safety standards for cars, toys or food, the duty of care applies basic standards for the online products we use every day," she said, and the duty "goes beyond social media" to chatbots, gaming platforms, apps and messaging services. There is no user-number or revenue threshold in the draft. It reaches carriage services, search engines, app stores, hosting providers, equipment suppliers and services that let users generate material with artificial intelligence.
The delegated powers are what the argument is about. Beyond the user empowerment tools provision, the draft lets the Minister determine by instrument what counts as seriously harmful material for all Australians, what counts as harmful to children, which design features have negative behavioural impacts, and which services are exempt. Each turns on what "the Minister is satisfied" may cause harm, a subjective test rather than an objective one. The draft already lists 12 categories of serious harm and five categories of harm to children, each ending in a catch-all.
Two constraints sit against those powers. Section 27B requires the Minister to seek the eSafety Commissioner's advice before making any of the instruments and to have regard to it. And the instruments would be disallowable, so the Senate could strike one down within 15 sitting days.
Wells did not dispute the scope of the power when it was put to her directly. "The rule-making power exists elsewhere in the Online Safety Act. I used it, you recall, to include YouTube in the social media minimum age laws when I became the Communications Minister," she said. She gave an example of how she might use the new one: consult the Commissioner, take her technical advice, "and then declare nudifying photos of teenagers in bikinis a new serious harm". On censorship, she said: "We can emphatically rule that out."
Shadow communications minister Sarah Henderson accepted the premise and rejected the method. "There is a legitimate debate about whether Australians should be able to switch algorithmic profiling on or off," she said. "But Canberra should not dictate what Australians see online by replacing one compulsory system with another." Big tech "should not be able to operate a black box which shapes what millions of Australians see every day without proper scrutiny", she said, and "parents should be in charge of how they protect their children online, not Silicon Valley and not Canberra". She also noted the timing. The draft landed on the day her own party was fighting a separate online safety enforcement bill in the Senate.
One Nation had moved four days before the draft was published. In a post on 4 September, Pauline Hanson said "Australia should be moving away from government censorship, not building the Orwellian machinery for it", and that the answer to social media's problems "is not giving a government like this the machinery to decide what is misinformation, what is dangerous and what Australians should be allowed to see, say and share". The party's formal statement on Wednesday was narrower, questioning whether the government could be trusted not to use the reforms for political advantage. Nationals leader Matt Canavan, also on 4 September, asked: "Who gets to define what's toxic? I start to get wary when we give the government power to make these decisions."
The duty of care was promised in November 2024, when then-communications minister Michelle Rowland told Parliament the government would legislate a recommendation of the Online Safety Act review, putting the onus on platforms so services "can't set and forget". The under-16 ban passed the following month, with penalties then capped at $49.5 million. The new draft roughly doubles that.
The consultation is short. Submissions close at midday on 22 September, a fortnight to respond to 75 pages, by email rather than through a public portal, and the government says the bill goes to Parliament this year. If it passes, the duty itself starts a year after royal assent. When Australians actually get the feed switch they were promised on Tuesday depends on an instrument that has not been drafted.
