This man has the job of working out how Australia’s social media ban will work

Since Prime Minister Anthony Albanese said he would pursue a world-leading social media ban for under-16s, one question has dogged his government: how will it work?

Tony Allen, who leads the company that will trial technologies for the Australian government, says there are three key options to assess users’ age that his company will be probing as it comes up with a toolkit for social media platforms.

The first tool Tony Allen and his team will test is age verification.Credit: Getty Images

Testing begins in the first half of next year on a sample group of about 1100 Australians, and a cultural adviser will ensure the specific needs of Australia’s Indigenous and immigrant communities are considered.

But no single tool would prevail, Allen said in an interview with this masthead. Nor will the government force social media companies to adopt any specific option, although Communications Minister Michelle Rowland has ruled out requiring people to upload proof of identity directly to social media platforms.

Rather, Allen’s company, the United Kingdom-based Age Check Certification Scheme, will give the government an assessment of how different tools will work in an Australian context. Social media platforms can then draw on that evidence to determine how they respond when the ban comes into play in 2026.

“If you’re trying to prove someone is a senior citizen to get a bus pass, or whether someone’s old enough to buy alcohol, different options will be better,” Allen said.

It’s the same with social media companies, which have different risks and audiences.

“If you think about LinkedIn and Discord, as two social media sites, their profile, risk and what they do are entirely different. So their solutions will be different,” he said.

The first tool Allen and his team will test is age verification.

“Put simply, [this] is where you find somebody’s date of birth and calculate how old they are,” he said.

It can be done by supplying a digital ID, driver’s licence, passport, bank records, school records or health records.

To test how well it works, Allen’s team will use more than 4000 legitimate, fake and altered identity documents from various regions.

They will gauge whether software can sort valid documents from those that are fraudulent or have been tampered with, and deal with worn, damaged or poorly scanned documents. They will also do security testing to ensure users can’t circumvent the system with deep fake images.

The second option in the toolkit is age estimation: tools that analyse biological or behavioural features that change with age.

“Our faces get older, our voices get deeper, our hands change the way they operate,” Allen said. Analysing these features – detecting wrinkles, for example – can indicate how old someone is.

Age estimation tools will be tested on Australians of different ages, ethnicities and genders to reflect the population. The company will trial static and real-time photos and use different angles and facial expressions. Voice, hand geometry and typing speed analysis will also be explored.

The third option is age inference: assuming someone’s age based on other information, such as their purchase history, browser behaviour or online activity.

“[It could look at] whether you have a mortgage, whether you have a .gov email address – from that you can infer someone is an adult rather than a child,” Allen said.

The trial will use avatar accounts that mimic personas and stimulate different behaviours – for example, by browsing child, teenage or adult-friendly content – to see how reliable the tools are.

There would be privacy concerns with all the options, Allen said, and “that’s a key part of our testing”.

Tech platforms will have to navigate other issues, such as whether the tools can be circumvented by VPNs that mask an internet user’s address.

Allen said VPNs were not difficult to spot because their addresses were often being accessed by thousands of people at a time, compared to a home address with a handful of users.

“A tech provider can look at that and say, ‘we need to do extra steps if a person is coming through a VPN’,” he said.

“The onus comes back to the providers.”

But it won’t be his job to tell the tech giants what to do.

“There won’t be one option that’s brilliant and everything else is rubbish. There’s risk that’s created in whatever scenario, and you need to find the right tools to address that risk,” Allen said.

The legislation passed last week allows a 12-month implementation period before companies must take action. Allen’s company will report to the government at the end of June, and the Senate by the end of September.

”We’ll give our assessed toolkit to the government, and it’s up to the government to decide what they do with it,” he said.

Cut through the noise of federal politics with news, views and expert analysis. Subscribers can sign up to our weekly Inside Politics newsletter.