An Australian government commissioned report has indicated that selfie-based age estimation software could facilitate a ban on underage users in social media, but also highlighted “unacceptable” levels of inaccuracy among certain demographics, prompting concerns ahead of the December rollout.

In a published report on Monday concluded that photo based age estimation tools are generally accurate, efficient and respectful of user privacy. However, it noted that accuracy diminishes for individuals close to the minimum age of 16.

The findings suggest a mixed impact when the ban is implemented: while Caucasians three years above the age threshold may face minimal disruptions, non Caucasian users and teenage girls could encounter significant challenges.

Social media platforms from Meta’s Instagram to Alphabet’s YouTube, will be required to take reasonable measures to block users under 16 or risk facing fines of up to A$49.5 million (approximately $32 million).

The trial further revealed high accuracy for users over 19, but those within three years of the cut-off-age fell into a “grey zone” characterised by greater uncertainity.

For instance, 16-year-olds has an 8.5% chance of being incorrectly assessed as underage, suggesting a need for additional verification methods such as ID checks or parental consents.

The reports also noted that while systems generally performed well across diverse groups accuracy decreased for older adults, non- Caucasian users, and female presenting individuals near the age threshold.

Media expert Justine Humphry from the University of Sydney expressed concerns about the feasibility of the ban within the tight timeline. “There are significant variations in accuracy, which is concerning with the impending implementation date”, she stated.

Communication Minister Anika Wells remarked that the report demonstrates the complexity of age assurance, emphasing that “there is no one size fit all solution”, but it highlighted the potential for effective options that protect user privacy.