The Identity Paradox: Reflections on AI from the boardroom and from the back of a taxi

Privileged and invigorated to have been part of lively debate this week, on the very topical thread of non-human identities, now proliferating across organisations as a consequence of the AI adoption surge. By 2027 it is predicted that 74% of organisations will be using AI Agents (Deloitte study, 2026).

The event was a #ChangeMakers round table on ‘The Security Shift’ that is consequently underway, as non-human identities proliferate at pace. The discussions progressed through the AI adoption curve, governance considerations, practical challenges and looking to ensure cross-org alignment, whilst the pace of change is more rapid than the process, procedures and guardrails that follow in its wake. I offered a point about the conundrum proposed by the ‘Identity Paradox’, whereby organisations experience intense downward exec pressure to implement AI at pace; but the balancing capabilities in controls and checks and balances are often stretched to keep up.

The quality of discussion in the room was excellent and perfectly on point for this critical topic - bringing together tech leaders operating where bleeding edge Identity, AI and Security intersect.  

With AI now being such a pervasive element of modern life, it was perhaps not a surprise that even the taxi ride home was ‘on topic’. Travelling back in a cab from Guildford station, I found myself in conversation with the driver, who was telling me his own AI story and the impact it had on the lives of both him and his family, particularly for his daughter, who was experiencing learning difficulties. Utilising AI to research and uncover local authority and other support resources, he relayed how AI was an accelerating force in helping him navigate the hugely complex routes to securing help for his daughter, via untangling a web of complex processes.  

That's the identity paradox in miniature. In the room, we debated how to govern non-human identities before they outpace our controls. In the cab home, AI was already, quietly, helping a father navigate a system built for humans, on behalf of a daughter who needed it most. Two very different rooms. The same question: can the systems we trust actually be trusted with what matters most to us?

AI and Identity discussions in the room