As enterprises race to adopt agentic AI, organizations are focusing heavily on the productivity gains autonomous software agents promise to deliver. But according to Entrust CEO Tony Ball, many have yet to grapple with a more fundamental challenge: establishing trust in a world where non-human actors increasingly perform work on behalf of people.
Speaking on the ID Talk podcast, recorded on-site at Identity Week Europe, Ball argued that the industry needs to think beyond authenticating human users and begin treating AI agents as identities that require governance, permissions and accountability.
“The agentic narrative we’re seeing right now is a massive accelerator to how we bring about a stronger trust layer, and how we put the right wrap around identity, both human and non-human,” Ball said.
Ball, who became Entrust’s CEO earlier this year after more than three decades in the identity and payments industries, described AI agents as an inevitable extension of enterprise workflows. Organizations are already discussing deploying agents to manage calendars, investment portfolios, strategic planning and countless other business functions, he noted. The challenge will be ensuring those agents operate within clearly defined boundaries.
“How are those agents going to be given privileges? How is an agent given a birth certificate, a death certificate? What privileged access does it have?” Ball said. “You need some type of check and balance to that. It cannot go unguardrailed.”
Rather than viewing agentic AI as simply another productivity technology, Ball argued that enterprises should begin thinking about identity infrastructure for autonomous software in much the same way they have historically managed human users.
That also means ensuring humans remain accountable for critical decisions. “How do humans stay at the center of the decision-making?” Ball asked.
As an example, Ball described a future enterprise environment in which employees are biometrically enrolled when they join an organization. AI agents could then perform routine work on their behalf, while high-risk actions – such as approving major financial transactions – would still require biometric confirmation from the human responsible for the agent. The approach, he suggested, would preserve many of the productivity benefits of autonomous systems while maintaining clear oversight during moments of elevated risk.
Ball also believes organizations should begin preparing internally now, even if their long-term AI strategies are still taking shape.
Entrust itself has established a program of 50 “AI champions” tasked with reverse-mentoring the company’s senior leadership on emerging AI capabilities. Ball said he personally spent three months working with a reverse mentor before assuming the CEO role.
“I’ve had a reverse mentor for three months in preparation to come into the role, because I need to understand how quickly this world is moving,” he said.
The initiative reflects Ball’s broader view that enterprises cannot afford to delay experimenting with AI, but neither can they deploy autonomous systems without governance. Instead, he recommends organizations identify promising use cases, establish AI governance teams, launch controlled pilot projects and gradually scale successful initiatives while maintaining appropriate oversight.
The rapid emergence of agentic AI, he argued, represents a shift that enterprises should approach with the same seriousness as previous technology revolutions, but with a stronger emphasis on trust. As software agents become increasingly capable of acting independently, Ball believes identity systems will need to evolve beyond verifying people alone, becoming the foundation for governing relationships between humans, AI agents and the organizations they serve.
Hear the full conversation via the media player below, or wherever you get your podcasts.
–
By the ID Tech Editorial Team
