Australia Just Told Big Tech: Let Users Kill the Algorithm

By TechPulse USA Newsroom — Canberra | 15 October 2026

On Thursday morning in Canberra, Communications Minister Anika Wells carried a piece of legislation into the Australian parliament that reads like a declaration of war on the defining feature of modern social media. The Digital Duty of Care Bill would give Australians a legal right to switch off the algorithms that decide what they see — and make the companies behind Facebook, TikTok and Instagram legally responsible for keeping under-18s away from pornography, online bullying, crime-glorifying content and material that encourages eating disorders. No other country has tried this exact combination.

What Canberra Announced on Thursday

Wells formally introduced the bill — the 2026 Online Safety Amendment (Digital Duty of Care) Bill — on the morning of October 15, following a September draft unveiling by Prime Minister Anthony Albanese. It does two things, and both target how platforms behave rather than what people post.

First, it forces the firms behind the biggest social platforms to give users a genuine choice: keep the algorithmic feed, or switch to one built only from accounts you follow. Australians over 16 could choose between algorithmically recommended content as their primary feed or a straightforward follow-based view. Second, it imposes minimum safety standards across social media platforms, gaming services, apps and AI chatbots — and holds companies directly responsible for shielding under-18s from pornography, online bullying, crime glorification, eating-disorder content, dangerous stunts and material promoting hostility toward women.

The enforcement teeth are real. Companies flouting the duty face fines of up to A$109.2 million — roughly US$75 million. The eSafety Commissioner would gain the power to remove AI-powered "nudification" services, and response deadlines for cyberbullying and online abuse complaints would be halved from 48 to 24 hours. We cross-checked the penalty figure against today's reporting before printing it: it comes from the minister's own official statement.

“If You Build It, You Are Responsible for Its Risks”

Communications Minister Anika Wells, who introduced the Digital Duty of Care Bill to parliament on October 15. The portrait shows her in front of the Australian flag. Photo: Department of Foreign Affairs and Trade / Sarah Hodges, CC BY 4.0, via Wikimedia Commons.

Wells did not tiptoe into the debate. At the National Press Club in Canberra — the venue Australian politicians reserve for the fights they actually want — she told the audience that "big tech has been allowed to live-test its products on Australians." Then came the sentence that will be quoted everywhere this bill is debated: "if you build it, you are responsible for its risks."

She went further. Wells said she believed Australians would one day look back on the country's minimum-age rules and think, "I cannot believe we used to let big tech have unfettered access to our children." And she reached for an analogy designed to make the opposition's deregulation argument look absurd: "We didn't abandon seatbelts because people didn't want to buckle up."

It was the most aggressive public statement an Australian minister has made about the industry — and it was carefully inoculated against the obvious counterattack. Wells insisted the bill targets platform conduct, not speech. To prove it, the introduced bill strips out a draft clause that would have let the communications minister personally define new categories of harmful content. Professional news material and entertainment media are explicitly exempt, and parliament gets scrutiny over any future harm categories.

The Algorithm Kill-Switch, in Practice

This is the part that should worry executives in Menlo Park and Singapore. The duty is not a vague suggestion to "do better." It requires a product change: a real, usable toggle between algorithmic recommendation and a chronological, follows-only feed.

It matters because the algorithm is the business model. TikTok's "For You" page, Instagram's Explore, Facebook's feed — these are persuasion machines tuned to maximise time-on-device, and they are the reason a teenager looking up one diet video can find herself drowning in eating-disorder content within a week. Canberra's theory is that the recommendation engine, not any single post, is the source of the harm. Give users the kill-switch, and you defuse the engine.

The engineering questions are real: what counts as an algorithm, legally? How prominently must the toggle be displayed? Lobbyists will spend heavily on those definitions in consultation. But the direction is unmistakable — Australia wants the default product to be one the user actually chose.

Penalties, Nudification Takedowns, and the 24-Hour Rule

Under the bill, platforms must proactively reduce risks — prevention, not complaint-response. The eSafety Commissioner gains the power to order the removal of AI-powered nudification services, the deepfake tools that generate sexualised images of real people and have become a scourge of schoolyards. Response deadlines for abuse complaints drop from 48 to 24 hours, and breaches carry penalties of up to A$109.2 million.

Wells has also said the government is considering letting individuals sue platforms for breaches of the duty — a private right of action not in the current bill, but dangled as the obvious next step if the industry drags its feet. That is the clause that should keep platform legal departments awake.

Three young women absorbed in their smartphones. Under the bill, platforms would be legally responsible for keeping under-18s away from pornography, bullying, crime-glorifying content and material encouraging eating disorders. Photo: Chris Olszewski, CC BY-SA 4.0, via Wikimedia Commons.

The industry's response so far has been silence. In September, when the draft was floated, Meta — owner of Facebook and Instagram — declined to comment, and TikTok and Snap did not respond to emails. No grand statement of cooperation, no preemptive product announcement. That reads like a holding pattern while lobbyists calculate their odds in the Senate.

The Under-16 Ban That Started All This

None of this arrived from nowhere. In December 2025, Australia rolled out the world's first social media ban for under-16s — a policy now being copied or studied by New Zealand, Britain, France and Turkey.

The experiment has had a rough year. Data from parental-control app Qustodio, reported in August, showed Australian under-16s actually increasing their use of Instagram and TikTok despite the ban. eSafety's March update touted around 5 million deactivated accounts — a number that drew a furious Senate Estimates grilling, with Senator Malcolm Roberts arguing it was padded with bots and dormant profiles and that eSafety could not prove how many were active child users. Estimates have put the share of under-16s still on the platforms at more than 80 per cent.

That record cuts two ways. The opposition calls it performative — big numbers, leaky enforcement. Wells' camp argues the ban was never enough, because age-gating does nothing about the algorithm everyone else still swims in. The duty of care is the answer to the ban's shortcomings. This is, after all, the same Australia that — as TechPulse USA reported — recently called out Valve over child-safety gaps on Steam: a country systematically testing every lever it has, from storefronts to feeds.

Europe Got Here First. Canberra Goes Further.

Brussels deserves some credit for the template. The EU's Digital Services Act already requires very large platforms to offer at least one recommender system not based on profiling — Article 38. TikTok and Instagram comply with non-personalised options, though as TechCrunch has noted, finding and switching to them takes active user effort; nobody made the safe feed the default.

But Australia's bill is a different animal. The DSA imposes transparency and risk-assessment obligations; the duty of care imposes an outcome obligation — your product must be safe by design, and if it is not, you pay. Analyst Ewan Lusty of the consultancy Flint Global called the move "a continuation of a trend rather than a completely novel innovation" — but the trend is now being driven from Canberra, not Brussels.

For American readers, the contrast is stark. While eight US states move against AI chatbots — as we covered in the eight-states-versus-OpenAI self-harm fight — and Washington dithers on a federal algorithm regime, Australia is simply doing it.

 

Our Take: This Is the Test Case for the Algorithmic Age

For a decade, platforms have had it both ways — telling advertisers the algorithm is the magic, telling regulators it is just a sorting tool. Canberra is calling the bluff. If the algorithm is the product, then the algorithm carries the duty.

Whether it works is a different question, and the odds are mixed. Labor needs the Greens to get this through the Senate, and the conservative opposition is vowing to fight it at every turn on free-speech grounds. The definitions fight over what counts as an "algorithmic feed" will be brutal. And the under-16 ban's leaky first year warns that legislation without enforcement is a press release.

But every concession the industry refuses in Canberra it will make eventually, everywhere. The US$18 billion Meta settlement in the United States, the EU's addiction probes, the countries copying Australia's age ban — the ratchet only turns one way. Wells' seatbelt analogy does real rhetorical work precisely because it is true. The platforms know it. Their silence this week is the sound of an industry deciding how much to lose, not whether to lose. If an Australian teenager can one day kill the algorithm with a single tap, the rest of the world will not be far behind — including Washington, where the debate over AI oversight is still stuck in first gear.

FAQ: The Digital Duty of Care, Answered

Does this mean every Australian can turn off the algorithm?

For users over 16, essentially yes. Covered platforms must offer a choice between an algorithmically recommended feed and a follow-only feed. The exact legal definition of an "algorithmic feed" and how visible the toggle must be will be fought over in consultation — but the obligation is in the legislation.

What happens to Facebook, TikTok and Instagram if they refuse?

Breaches carry fines of up to A$109.2 million — roughly US$75 million. The eSafety Commissioner would also gain new powers, including removing AI nudification services, and Wells has floated letting individuals sue platforms directly for breaches — a step not yet in the bill but clearly on the table.

Is this the same thing as Australia's under-16 social media ban?

No, but they are siblings. The ban, in force since December 2025, blocks children from social media accounts outright. The duty of care targets the machinery underneath — the recommendation engines and safety practices affecting everyone, with extra protections for under-18s. Canberra sees it as the follow-up to a ban that leaks.

What about free speech?

The opposition says the bill hands the government too much power to control speech and vows to fight it in the Senate. Wells' answer: the law regulates what platforms do, not what people say. The clause letting the minister define new harm categories was removed, news and entertainment are exempt, and parliament gets a say over future definitions. The Senate fight over that line decides the bill's final shape.

Could this happen in the US?

Not soon at the federal level — Washington has no equivalent duty-of-care statute for recommender systems, and a First Amendment challenge would be immediate. The realistic American path is the one already visible: state attorneys general extracting settlements that force product changes, as with Meta's US$18 billion deal last month. Canberra's experiment is the live test of whether legislation can go further.

Sources: Agence France-Presse reporting via The Straits Times, The Standard and Free Malaysia Today; MLex legislative analysis; The Express Tribune; Al Jazeera; The Taipei Times/Bloomberg; TechCrunch on the EU Digital Services Act. Bill details cross-checked against Communications Minister Anika Wells' official statement.

TechPulse USA is a newsroom tracking how artificial intelligence is actually being bought, sold and deployed — not just announced.