AI-powered selfie age verification booms as governments enforce online restrictions, with Yoti processing one million daily checks for clients like Meta and TikTok
LONDON: AI-powered age verification using selfies is experiencing rapid growth as governments worldwide impose stricter online controls.
The technology offers a fast method to enforce regulations like Australia’s upcoming social media ban for users under 16, effective from December 10.
Users simply take a front-facing selfie with their phone or computer camera, receiving an automated age determination within one minute.
Roblox gaming platform displays pop-up messages like “We estimated your age is 18 or older” after analysis.
At Yoti’s London headquarters, testing involves mannequin heads equipped with wigs and masks to challenge the system’s accuracy.
Their AI consistently identifies these as non-human faces, stating “We can’t be sure that the image was of a real face” during demonstrations.
Yoti CEO Robin Tombs explained that “the algorithm got very good at looking at patterns and working out, ‘this face with these patterns looks like a 17-year-old or a 28-year-old'”.
The startup now processes approximately one million age checks daily for major clients including Meta, TikTok, Sony and Pinterest.
Yoti achieved profitability this year after generating £20 million in revenue during the 12 months to March.
The company projects a 50% sales increase in its current financial year.
Thirty-four competitors including Persona, Kids Web Services, K-id and VerifyMy belong to the Age Verification Providers Association.
AVPA previously forecast nearly $10 billion annual revenue for the sector across OECD nations by 2031-2036, though recent projections remain unavailable.
AVPA director Iain Corby remains cautious about future predictions, noting “There’s so much can happen. Regulation is moving very fast, technology is moving very fast”.
Privacy concerns emerge as cybersecurity professor Olivier Blazy warns the tools are “potentially a bit intrusive or dangerous to people’s private life”.
Blazy suggested “the ecosystem is tilted towards AI at the moment, but there may be a pull-back in the coming years to better protect users’ private life”.
System vulnerabilities include makeup manipulation, with Blazy noting “just with normal makeup, it’s relatively easy to make yourself look older or younger”.
Algorithmic bias presents another challenge, particularly with non-white faces where accuracy can decrease.
An Australian report found “underrepresentation of Indigenous populations remains a challenge” that vendors are addressing.
Tombs acknowledges limited training data for certain age groups and skin colours but insists their tool detects “false accessories or they are using makeup and other such things”.
Yoti immediately deletes all data after analysis to address privacy concerns.
Platforms can adjust verification thresholds, often requiring systems to judge visitors as over 21 for under-18 restrictions.
Users in uncertain age ranges may need traditional verification methods like scanning official identification documents. – AFP
