LONDON – In the latest update to the organization’s draft guidelines for implementing the UK’s Online Safety Act, UK communications services regulator Ofcom appears to suggest that artificial intelligence will be part of the technology used to verify internet users’ age for the purposes of restricting access to online pornography and other content deemed harmful to minors.
In a document titled “Implementing the Online Safety Act: Protecting children from online pornography,” Ofcom “sets out guidance on highly effective age checks to stop children accessing online porn services,” listing methods which include “photo ID matching, facial age estimation and credit card checks.”
“Under the Online Safety Act, sites and apps that display or publish pornographic content must ensure that children are not normally able to encounter pornography on their service,” Ofcom states in the document. “To do this, they must introduce ‘age assurance’ – through age verification, age estimation or a combination of both – which is ‘highly effective’ at correctly determining whether a user is a child or not. Effective access controls should prevent children from encountering pornographic content on that service.”
Later in the document, Ofcom details what the organization considers “highly effective methods of age assurance,” offering a “non-exhaustive list” of methods “that we currently consider could be highly effective.” Included in the list are “open banking,” “photo identification matching,” “facial age estimation,” “mobile network operator age checks,” “credit cards checks,” and “digital identity wallets.”
The implication that AI could be a component of the technology employed by the Ofcom system comes in the context of a footnote to the summary of how “facial age estimation” might work under the system.
While the summary merely states that the “features of a user’s face are analyzed to estimate their age,” the footnote offers more detail about what this means.
“Our draft guidance also suggests that a ‘challenge age’ could be set,” the document stats. “This could mean where the technology estimates the users’ age to be under 25, for example, that user would undergo a second age-check via an alternative method.”
Ofcom adds that the organization is “aware that a wide range of age estimation methods exist.”
“At present, we have only proposed including facial age estimation in our guidance, as we do not have evidence to suggest that other methods of age estimation are currently capable of being highly effective, are sufficiently mature technologies, or are being deployed at scale,” the footnote continues. “We will continue to review this position over time as technologies evolve.”
In other words, what Ofcom has said is that AI might, at some point in the future, be a part of the technology used to “assure” (rather than “verify”) a given user’s age.
The guidance also makes clear that “weaker” forms of age assurance currently common among adult sites “won’t be enough.” These insufficient methods include self-declaration of age; online payment methods which don’t require a person to be 18” and “general terms, disclaimers or warnings.”
“Regardless of their approach, we expect all services to offer robust protection to children from stumbling across pornography, and also to take care that privacy rights and freedoms for adults to access legal content are safeguarded,” said Ofcom Chief Executive Dame Melanie Dawes.
The full details of Ofcom’s proposal spawn hundreds of pages across multiple documents – and there’s a great deal more in the mix than just rules which pertain to adult entertainment sites and apps. Many forms of content which aren’t remotely “pornographic” will also be subject to the Ofcom rules, including video content that “might impair the physical, mental or moral development of under-18s,” or “are likely to incite violence or hatred against particular groups,” or “directly or indirectly encourage acts of terrorism;” or “show or involve conduct that incites racism or xenophobia.”
Whatever one might think of some of those forms of expression listed above, a good deal of it is likely protected under the First Amendment in the United States, which, contrary to popular belief, does not prohibit expression commonly referred to as “hate speech” – presumably meaning that UK users will be denied access to U.S.-based sites which offer such content without requiring users to satisfy a “highly effective” age assurance measure.
Whatever form the final Ofcom rules take, they will not be implemented any time soon. According to Ofcom, the organization expects to publish its final guidance in “early 2025.”
