Technologies to apply the prohibition of social media from the Australian government for those under 16 are “private, robust and effective”. It is according to the Preliminary results of a communicated federal government trial that has almost finished testing them.
The conclusions, published today, can give the government greater confidence to move forward with the ban, despite a Continuation of expert criticism. They could also mitigate some of the concerns of the Australian population concerning the privacy and security of the implications of the ban, which should start in December.
For example, a report based on a survey of nearly 4,000 people and Released by the government earlier this week Found nine out of ten supports the idea of a ban. But also found a large number of people were “very concerned” of the way in which the prohibition would be implemented. Almost 80% of respondents had confidentiality and security problems, while about half of the concerns about the precision of age insurance and government surveillance.
The preliminary tests of the trial brush a pink table of the potential for technologies available to check the age of people. However, they contain very few details on specific technologies and seem to be in contradiction with what we know about the age insurance technology of other sources.
From facial recognition to the recognition of hand movements
The ban on social media for under 16s was legislated in December 2024. amendment the law obliges technological companies to provide “Alternative age insurance methods“For account holders to confirm their age, rather than counting solely on an identity document issued by the government.
The Australian government ordered an independent trial To assess the “Efficiency, maturity and preparation for use“Of these alternative methods.
The test is led by the Check age certification scheme – a company based in the United Kingdom specializing in tests and certification of identity verification systems. He understands 53 sellers who offer a range of age insurance technologies To guess people’s age, using techniques such as facial recognition and hand movement recognition.
According to the preliminary conclusions of the trial, “age insurance can be carried out in Australia”.
The trial project director, Tony Allen, said “There are no significant technological obstacles” to people’s age insurance online. He added that solutions are “technically possible, can be flexibly integrated into existing services and can support the security and rights of online children”.
However, these statements are difficult to compete with other evidence.
High error rate
ABC yesterday reported that the trial had found facial scanning technologies ”poorly identified several times“Children as young as 15 are in their twenties and thirties. These tools could only guess the age of children “in an 18 -month -old range in 85% of cases”. This means that a 14 -year -old child could have access to a social media account, while a 17 -year -old could be blocked.
This complies with the results of global testing of scanning technologies carried out for more than a decade.
A Series of studies in progress Age estimation technology by the National Institute of States and Technology of the United States shows algorithms “fail considerably when you try to differentiate minors“Of various ages.
The tests also show that error rates are higher for young women than for young men. Error rates are also higher for people with darker skin tones.
These studies Show that even the best age estimate software currently available – Yoti – has an average error of 1.0 years. Other software options confuse the age of someone 3.1 years on average.
This means, at best, a 16 -year -old could be estimated at 15 or 17 years; At worst, they could be considered 13 or 19 years old. These error rates mean that a significant number of children under the age of 16 could access social media accounts despite a ban in place, while some more than 16 could be blocked.
Yoti also explains companies that need To check the exact ages (like 18) can set higher age thresholds (like 25), so fewer people under the age of 18 go through age control.
This approach would be similar to that taken in Australia retail alcohol sectorWhere the sales staff check the identity card for anyone who seems to be under 25 years of age. However, many young people have no identity issued by the government required for an additional age check.
It should also be recalled that in August 2023, the The Australian government has recognized That the age insurance technology market was “immature” and could not yet meet key requirements, such as work reliably without bypass and balance confidentiality and security.
Outstanding questions
We do not yet know exactly what methods of methods will use to check the ages of account holders. Although facial sweeping technologies are often discussed, they could use Other methods To confirm age. The government trial has also tested voice and hand movements to guess the age of young people. But these methods also have precision problems.
And we do not yet know what people will have if their age is poorly identified. Will parents be able to complain if children under the age of 16 will have access to accounts, despite the restrictions? Can older Australians who are incorrectly blocked can appeal? And if so, to whom?
There are other outstanding questions. What prevents someone who is under 16 from asking someone over 16 to create an account on his behalf? To alleviate this risk, the government could demand that all social media users check their age at regular intervals.
It is also not known what error in age estimation, the government can be willing to accept in the implementation of a ban on social media. THE legislation Said that technological companies must demonstrate that they have taken “reasonable measures” to prevent those under 16 from having social media accounts. What is considered “reasonable” has not yet been clearly defined.
Australians will have to wait later this year that the complete results of the government trial will be published and know how technological companies will react. With less than six months before the entry into force of the ban, social media users still do not have all the answers they need.