Artificial Intelligence (AI)

Identity and access management in artificial intelligence. So we see artificial intelligence used in some biometric services already. A Chinese businesswoman was accused of jaywalking after an artificial intelligence camera spotted her face on an advert for a bus. This is a camera based system, automated, using AI to match biometric templates. UEBA is a term coined by Gartner, User and Entity Behaviour Analytics. Big data platforms are increasingly using user behaviour analytics by the analysis of huge amounts of data. And this allows the detection of, or this supports, the detection of insider threats and advanced persistent threats. We start to see things like keyboard analytics in use as part of an authorization process. So you authenticate to a solution.

You have an authorization model in place. And some of these AI systems now can start to learn heuristically, over time, how you use a keyboard. And if you type in a different way, we will type slightly differently. If you type in a different way, it flags a concern. So this kind of adaptive pattern based approach, this pattern recognition, is increasing massively. Machine learning and artificial intelligence, very, very strong development areas at the moment. So these provide a form of ongoing authorization that is process based on anomaly detection. For these to be effective, they have to build a profile of behaviour first, either based on a standard user, or more commonly based on individual user patterns.

And you do have the potential for disruption through false positives, as we said, at the start, the identification of a poster rather than actually of the individual. So things like face recognition, and with internet of things and mobile voice recognition, and adaptive authentication, all of these become much more fluid as an authentication and an authorization model. We see over time this developing and providing a much more, much more dynamic approach for each of those services, authentication and authorization.