Age Assurance = Age Verification + Age Estimation
The UK Government and its regulators increasing refer to the concept of “Age Assurance”. This has caused some confusion, so we set out below our understanding of this term and how it is distinguished from the more widely used term “Age Verification”.
Age Assurance is the broadest term for methods to discern the age or age-range of an online user.
Age Verification is a subset of Age Assurance, providing higher levels of confidence in the age or age-range of a user.
Age Estimation is also a subset of Age Assurance, but providing lower levels of confidence in the age or age-range of a user, often relying on artificial intelligence and machine learning techniques.
A UK government official describes Age Verification as “the gold standard of Age Assurance.” The BSI defines it as:
2.1.8 age verification
determination of an individual’s age involving a full identity verification process
(Source: PAS 1296, see below)
This definition, however, has been overtaken by technology, with methods of age verification now available that do not require full identity verification process.
Typically, Age Verification is confirming age or age-range to a sufficient standard to comply with laws and regulations that specify a particular minimum age which must be checked exactly e.g. 21, 18 or 13
Age Estimation methods which do not meet the standards required for Age Verification will typically only indicate a likely age range and do so with a lower level of statistical certainty than Age Verification.
Examples of lower level age estimation techniques given by the UK Government so far typical include the use of artificial intelligence. The Secretary of State cited analysis of the typing style and speed of the user as one method which could be developed.
Age Verification is still possible using artificial intelligence. For example, facial analysis can provide a very high level of confidence that a user is over 18, if the software tests for an age above that e.g. 23. Systems can be audited to demonstrate they would correctly confirm a user was over 18 99.99% of the time, if they only did this when the estimate indicated the user was over 23. This is a far higher standard of accuracy than a human estimating the age of a customer in person, so satisfies regulators in most situations.
Age estimation does not include very simple methods such as ticking a box to confirm age or entering a date of birth. These data points could, however, be used in conjunction with other methods that would provide additional evidence to confirm age to a required level of confidence to be considered as Age Estimation.
Who defines these terms?
The British Standards Institution has defined a Publicly Available Specification, BSI PAS 1296:2018 for “Online age checking. Provision and use of online age check services. Code of practice.” This sets out the basis for describing the levels of assurance offered by different methods of age checking.
This standard will be the basis of a new international ISO standard, developed during 2021, with UK Government sponsorship, which include explicit reference to Age Assurance.
For more information on our work developing standards, see ……