Australia’s Feed Prompts Leave the Algorithm Switched On

Australia released draft Digital Duty of Care laws on 8 September 2026 that would make social media platforms ask users to pick a default feed. Fines for a breach run up to A$109.2 million, and the eSafety Commissioner would police the rules.

The slogan is My Feed, My Way. The design still leaves personalised ranking as the easy home screen unless someone chooses the following feed instead.

Platforms Must Ask, but the Algorithm Stays On

The Albanese government published the exposure draft of the Online Safety Amendment (Digital Duty of Care) Bill 2026 the same day it staged a Canberra press conference with parents and youth charities. Communications Minister Anika Wells said the duty is meant to work like basic safety rules for cars, toys or food, and that it is not about controlling adults online.

The feed piece is narrower than the branding. Platforms would have to send a notification to new and existing users offering a choice over the default feed. Users could opt in to personalised content ranked by an algorithm, or opt out and see friends and creators they already follow.

Wells told the press conference that platforms must give people over 16 a choice on what feed they see when they open the app, and then respect that choice. That is a forced ask, not a buried toggle. It is also not the Greens’ version, which would leave ranking off until someone turns it on.

This is not about giving government control, it is about giving people control.

Anthony Albanese, Prime Minister, media release, 8 September 2026

Prime Minister Anthony Albanese called the package sensible and said companies that ignore it would face penalties of more than A$100 million. The written release is more precise: failure to comply with the duty could attract penalties of up to A$109.2 million.

Instagram, Facebook and X already ship a Following tab. The extra move in this draft is to put the choice in a notice, then make ignoring that notice a compliance problem. The ranking engine that fills a For You page can stay exactly where it is if the user taps the familiar option.

What a Following Feed Does to Time Spent

Platforms have a measured reason to keep that familiar option close. In 2023, Andrew Guess of Princeton and colleagues reported in Science on a field test that moved consenting Facebook and Instagram users onto reverse-chronological feeds during the 2020 US election. Time on the apps fell. Activity fell with it.

Guess later put the drop at about 20 percent less time on Facebook and about 11 percent less on Instagram. The study did not find a clear shift in polarisation or political knowledge over three months. It did find that people spent the spare attention somewhere else.

CHRONOLOGICAL FEED, 2020 FIELD TEST

On Facebook, the chronological diet also raised the share of posts from sources the researchers marked untrustworthy, from 2.6 percent to 4.4 percent. Uncivil posts went the other way. The ranking engine was not a simple harm pump, and turning it off was not a simple clean-up.

A following feed is quieter. It is also easier to abandon. That is the bind inside My Feed, My Way. Canberra can count whether a notice appeared. It cannot count whether anyone wanted the poorer product on the other side of the tap.

Europe Already Offered the Switch, Quietly

The European Union’s Digital Services Act, in force for the largest platforms since 2023, already requires a recommender option that is not based on profiling. That is the same substantive right Australia is writing into domestic law. The DSA does not make anyone keep putting that option on the home screen.

Take-up has been modest, which is the point of the Australian notice. In October 2025 a Dutch court held that Meta had made the non-profiled timeline on Facebook and Instagram too hard to keep, because the ranked feed kept returning. The judgment treated a buried setting as a setting that does not count.

Australia is trying to close that gap with a prompt. The prompt still offers the ranked feed as a valid answer. If the Science experiment is a guide, many people will pick it, then open TikTok anyway when the following list feels slow. The legal win is a recorded choice. The product win stays with the ranking model that already pays for the service.

The Age Ban Left Most Teens on the Apps

The draft lands nine months after Australia’s under-16 social media ban took effect on 10 December 2025. Albanese said that action is working, that more than 5 million accounts have been removed or deactivated, and that more than 20 countries are taking similar steps. He also said the ban was never going to be perfect.

eSafety’s own three-month evaluation, released on 30 July 2026, is cooler. Among under-16s, account ownership fell to 42.1 percent from 52.4 percent. Use of an age-restricted platform, with or without an account, fell only from 85.9 percent to 81.5 percent. Most children who had accounts before the start date still had one, or had opened a new one, and weak age checks were the main reason given.

THE BAN IN THE FIRST MONTHS

• Accounts cut: Albanese said more than 5 million under-16 accounts were removed or deactivated.
• Ownership: eSafety’s July evaluation recorded a drop from 52.4 percent to 42.1 percent.
• Still using the apps: Reported use fell from 85.9 percent to 81.5 percent.
• Open files: From 31 March 2026, eSafety has been investigating Facebook, Instagram, Snapchat, TikTok and YouTube.

On 31 March 2026, eSafety Commissioner Julie Inman Grant said the office was moving into an enforcement stance after finding poor practice at those five platforms for possible non-compliance. Civil penalties under the age rule then sat at up to A$49.5 million. In June 2026 the government moved to double that cap to A$99 million and to widen eSafety’s power to pull internal files.

PRACTICES ESAFETY FLAGGED IN MARCH

• Wrong prompt: Children whose declared age was already under 16 were still pushed into age checks.
• Repeat tries: Under-16s could run the same check again until it returned a 16-plus result.
• Weak reporting: Paths to flag an under-age account were poor or hard to use.
• New accounts: Measures against fresh under-16 sign-ups were judged insufficient.

Age is hard to prove. A notice on a feed is not. Either the prompt is there at the stated moment, or it is not. For a regulator that has spent 2026 arguing about selfies and workarounds, that is a cleaner file. It is also a file that can be green while the ranked feed, and the time it buys, barely move.

Chatbots, Games and a List of Named Harms

The feed notice is the poster. The duty underneath is wider. Digital services, including online games, apps and AI chatbots, would have to protect under-18s from design features with negative behavioural effects, including addictive features and tools that can hit a young person’s self-esteem. Platforms would also have to write down the steps they take against identified risks and keep those steps working over time.

Wells said tech firms can still move fast, but they will no longer be allowed to break things. She pointed to girls getting chatbot advice on surviving on 600 calories a day, boys being bullied by strangers in games, and young people copying dangerous stunts they saw online. Jim Hungerford of the Butterfly Foundation, standing with the ministers, said every month Australia loses a hundred people to eating disorders, and that the helpline hears daily from people whose feeds are packed with weight-focused posts.

NAMED HARMS FOR UNDER-18S

• Eating disorders: Content that promotes, encourages or accelerates them.
• Misogyny: Content that promotes hostile ideas about women and gender equality.
• Pornography: Listed as a harm the duty would require services to block from under-18s.
• Crime and stunts: Content that glorifies crime or life-endangering stunts.
• Distress: Content that causes serious mental health distress, including abuse and bullying.

For adults, Wells said the hard stops are illegal acts: child sexual exploitation and grooming, violence and sexual violence, menacing or harassing conduct, and the promotion of terrorism and extremism. Apart from that, she said, adults get a choice of default feed. The draft would also let eSafety issue removal notices for nudify apps or sites, and speed up the child cyber-bullying and adult cyber-abuse schemes.

That is a systems duty, not a takedown queue. It is also the part that does not fit on a pop-up. A following feed will not, by itself, keep disordered-eating videos out of a chatbot, or stop a game from running a streak mechanic. The notice is auditable. The rest of the duty will be argued in risk papers, product changes and, if eSafety follows the age-ban path, another round of information notices.

Who Gets to Name a New Harm?

The Coalition has not landed on one line. Shadow communications minister Sarah Henderson has attacked recommender systems as psychological manipulation of children, and has also called the proposed powers a serious threat to free speech. Opposition leader Angus Taylor said he was deeply sceptical and feared an attempt to censor social media. Shadow treasurer Tim Wilson has raised similar worries about restricting opinion.

Wells and Albanese spent the press conference trying to kill that reading. Asked whether harm could cover political commentary, Wells said the adult prohibitions are the illegal list, and that the rest is choice: the algorithm, or the following feed of friends and chosen creators. Albanese said people should read the draft and talk to the parents in the room before they call it a muzzle.

The Greens, from the other side, want a harder default. Their Fix Our Feeds bill, already in the Senate, would let users drop recommended content at any time and build a duty of care around large providers. Safety groups have pushed for ranking to stay off until someone opts in. The government draft takes the softer path and tries to make up the difference with a notice.

That split is the one that will matter in the Senate. A following-feed toggle is already on the major apps, which is why the argument on the day of the launch slid so quickly onto who can add a new category of harm and how fast the Senate can disallow it. The slogan is choice. The live fight is the catalogue.

The department has asked digital platforms, industry bodies, civil society groups and advocates for feedback by 22 September, with comments due by 12pm that Tuesday, and says a bill will go to parliament later in 2026. The mailbox closes 14 days after the draft went public. The ranked feed, unless enough users tap the other option, stays where it is.