By Grace Crivellaro in Canberra

Concerns are mounting tech giants will wrongly identify kids as adults when age verification checks begin to rollout on social media platforms in a world-leading ban for under-16s.

The social media ban will begin on Wednesday and includes platforms YouTube, X, Facebook, Reddit and Threads, with the government stating more may be added at a later stage.

Many platforms captured under the eSafety commissioner’s list of banned platforms will use facial scanning as a first step to age assurance.

It is up to platforms to decide what age verification methods to use.

The only requirement is that requesting ID cannot be the only form of checking, the government has said.

Some platforms such as TikTok say they will use facial-age estimation service Yoti, with a government trial revealing it has good accuracy overall, but 34 to 57 per cent of 13 to 15 year olds were falsely classified as over 16.

The report also found 55 per cent of 18 year olds were classified as under 18.

The federal opposition says the government is relying on “imperfect technology”.

“The government set the (December 10) deadline, there were errors in the trial, but they continued with the deadline,” communications spokesperson Melissa McIntosh told AAP.

Snapchat has said users who have been wrongly flagged as being under 16 can appeal by going through a bank card check or providing government ID such as a passport or driver’s licence.

Ms McIntosh holds concerns over age assurance companies storing data offshore.

“The best answer I’ve been able to receive from platforms (is that personal data will be deleted) as quick as possible,” she said.

“Nobody can say what quickly as possible actually means.”

Communications Minister Anika Wells has said the ban’s rollout will not be perfect, but will ultimately shift the culture and protect the seven out of 10 children who witness online harm.

RMIT University’s Centre for Human-AI Information Environments director Lisa Given said there were risks with the age-assurance measures.

A 15-year-old could be misidentified as over 16 and would retain access to their accounts, while older teens could be banned, she said.

“These technologies have typical error rates between one to three years,” Professor Given said.

“This means a 17-year-old may be blocked from accessing their account, if the technology estimates them to be younger than 16, and be required to provide evidence of their real age.”

News all day, every day at CityNews.com.au.