澳洲擬立法要求社交平台提供推薦演算法關閉選項

Toxic algorithms wouldn't let them go - now Australia wants an off switch

ByTiffanie Turnbull

Sydney

Warning: This article contains an account of living with an eating disorder

Varsha Yajman was already self-conscious about the way she looked when she joined social media as a 13-year-old.

Endless workout videos making impossible promises filled her feeds. Through the screen, girls she'd never met urged her to sign up to dangerous calorie challenges. Others proudly promoted their height and weight - even the number of times they had been sent to hospital - on their profiles in a form of twisted competition.

Every night as she fell asleep, Varsha would watch a stranger on the internet diarising what they'd consumed that day.

"I was like, 'I don't need to eat. I can just watch other people eat'," Varsha tells the BBC.

"I would find that I'd be down these really deep rabbit holes... it was like no matter how hard you tried to get out, you couldn't."

Now 24 and in recovery from anorexia, Varsha says she is still served weight loss content on social media - despite having every safety setting turned on, and an exhaustive list of banned words.

It is people like her the Australian government says it is trying to protect with a new suite of laws cracking down on big tech, which were introduced to parliament this week.

The centrepiece of their proposed reforms: an algorithm switch which would force social media platforms to let users turn off recommended content. The idea is that you would only see posts from accounts you follow.

It isn't the first government to try this, but existing algorithm controls are often transient and come with trade-offs in app functionality.

Australia has set out to make its legislation tighter to stop this. And it hopes its new laws – which would also impose a legal duty of care on online services – will help Australians become happier and healthier.

But critics are concerned Prime Minister Anthony Albanese's government has bitten off more than it can chew in its fight with big tech, while others, including the Trump administration, say the reforms threaten free speech.

The choice

Flynn Martin knows what it is like to be sucked into an algorithmic spiral.

Self-proclaimed misogynist influencer Andrew Tate and other manosphere figures started overwhelming the then-20-year-old's social media feeds about four years ago.

It was a time when Flynn, who was eventually diagnosed with bipolar disorder, was insecure, frustrated with his life and struggling with his mental health. He says he fell for the rhetoric.

He soon went viral in Australia for two videos: one ranting about whether it was OK to be a straight white male, and another where he criticised a date for her frozen yoghurt order.

"You find yourself in like a bit of an echo chamber... It's easy to get swept up in that," he tells the BBC.

"I really bought into the idea we were living in this matrix, we had this really negative system that was designed against you. And the only way to break out of it was becoming this alpha male character."

The Melbournian was only pulled from the manosphere after a period of psychosis and a hospital stay.

Nowadays, he finds it tough to look back at that version of himself: "While there's a bit of shame and embarrassment... I think there is a bit of empathy for that younger man."

Both Flynn and Varsha know social media alone isn't to blame for the worst chapter of their lives, but say it was a key factor.

Countless others have stories of being confronted by social media content they didn't seek out, and didn't want to see – and there's long been suspicions the algorithms curating these platforms are harming users.

Social media firms argue the potential risks are being overstated and that they already provide users with effective safety tools.

But research emerging from around the world has been sketching a different picture.

Studies have found social media use triggers a similar brain response to gambling and substance addiction, external; that algorithms – often unsolicited – serve up content which harms users' mental health, external, and that they're spreading disinformation and increasing polarisation, driving violence offline.

Court cases and whistleblowers have aired claims that these companies know the dangers and are willing to overlook them as this technology lifts user engagement and their bottom line.

"The result is that they promote content that is outrageous or that invokes strong emotions from people - so it might be anger or fear or desire or lust or insecurity, whatever it is, it's going to keep us scrolling," Chanel Contos tells the BBC.

While studying the topic at university, the activist - who helped reform sexual consent education standards in Australia - came up with a campaign for change.

"Fix Our Feeds" demanded recommended algorithms be turned off by default, and social media users be given the choice to opt into them.

Contos envisaged a nostalgic return to the social media of the early 2010s – a time when a person's social media was filled with posts from friends or people they actually followed and their feeds were finite.

More than 15,000 people signed her petition, and the campaign also won the backing of two former prime ministers, several MPs, influential figures, mental health advocates, and online safety experts.

The government in September said it had listened to the Australians who had backed Contos's campaign, and announced a version of her proposal: forcing big tech to introduce an algorithm switch - but stopping short of requiring it be automatically set to off.

"This is not about giving government control, it is about giving people control," Albanese said.

"If people want to doom scroll… that's fine, that's up to them."

The policy is part of the government's Digital Duty of Care bill, which would require online services – social media firms, gaming sites, chatbots and more – to "ensure, so far as is reasonably practicable, a safe online environment".

Children must be protected from "harmful" material and design features – for example, pornography and material which encourages eating disorders or harmful attitudes towards women – and adults must be protected from "seriously harmful" material and conduct, a much narrower list of illegal content. Breaches of the law could attract fines of up A$100m (£52m, $70m).

"The digital duty of care answers a lot of the 'what abouts' of the social media minimum age law because it's much broader; it captures all online services and it shifts the onus of responsibility from the shoulders of victims onto the shoulders of the big tech operators who seek to make money off Australian users," Communications Minister Anika Wells told the BBC.

The challenges

There's been a muted response from big tech. Meta, Snap, TikTok, X and Google, which owns YouTube, have all declined BBC requests for comment, and have not made public statements about the legislation.

In the days after the announcement, Instagram boss Adam Mosseri argued non-algorithmic feeds would mean users saw fewer posts from friends, and said he didn't believe his app was addictive – although a US jury recently ruled otherwise.

"I think we conflate the way we colloquially use the word 'addicted' – like, 'I'm addicted to this show, and I'm binge-watching it' – and clinical addiction," he told a press briefing in Sydney.

Right-wing parties in Australia – and the Trump administration – have said the algorithms proposal and the broader bill both risk forcing over-moderation and censorship.

But those who defend the law say that won't happen because you can still follow who you want and share what you want. "If adults listen to manosphere podcasts, or people post their views about the government, or make memes, or want to say the moon landing was fake – they can continue to do so without restriction," Wells said this week.

The proposal seems to be far less divisive amongst the public and online safety experts compared with Australia's social media ban for children.

However, the big question is: how will Australia implement this policy when, so far, it has been struggling to implement the ban? Surveys suggest most children who were on social media before the ban still are, and no fines have been issued.

The EU's attempt at introducing algorithm controls in 2023 – after which platforms rolled out following-only feeds – is a cautionary tale.

Social media firms have indicated the take-up of safety and control features like this has been low and cite it as evidence users like their products the way they are. Experts say there is a reason for that.

"Platforms do what the law orders, but in such a clunky way that users don't actually enjoy the option and stick with the default, algorithmic experience," researcher Chris Zomer, from the ARC Centre of Excellence for the Digital Child, wrote in The Conversation, external.

On Instagram for example, when you turn off the algorithm, you can't see direct messages or access your profile and settings. When you leave the app, it reverts back to an algorithm-governed feed on your return.

The only place where these issues don't exist is the Netherlands. It was changed there after a Dutch court last year found Meta in breach of the EU's Digital Services Act.

The rules for Australia aren't set in stone yet, but the government says following-only feeds must not be harder to use or less functional, that algorithm settings must be "obvious", and that they must not change without an "explicit choice" by the user, even if they leave the app.

Buoyed by her success in Australia, Contos is now taking her campaign global.

"We feel like Canada, UK, Australia are all very aligned on online safety... Then if those three jurisdictions are doing it, it could become a critical mass thing."

Though experiences like theirs helped inspire these reforms, both Flynn and Varsha are somewhat sceptical of them.

Varsha says she struggles to believe a country like Australia can outmuscle big tech: so she's not sure the Digital Duty of Care bill will work. And Flynn says lawmakers should be "realistic" about the potential of the algorithm switch: "At the end of the day we use social media as a form of entertainment."

Polling earlier this month indicated two-thirds of their fellow Gen-Zs would keep their social media algorithms – though nearly half of Australians more generally would turn theirs off.

Though Varsha admits she is unlikely to make much use of an algorithm switch, she appreciates having the choice.

"On a bad day I'm still susceptible to all of the things that I was at 17. To know that, if I'm having a bad day, I can opt out makes me feel really good."

If you've been affected by the issues raised in this story, you could speak to a health professional or an organisation that offers support. Details of organisations offering advice and support for people in the UK are available from BBC Action Line.