Identity Standards for AI | CIAM, OAuth & Agentic Identity

In this first episode, host Joe Miller, Partner Sales Manager for the Americas on the Thales CIAM team, is joined by identity experts Nate Szytel, CEO of Next Reason, and James Bonifield, co-founder and CEO of Anthropic Identity, for a candid conversation on identity standards for AI: “déjà vu all over again.”

AI is creating a new identity crisis. In the past, major technology shifts pushed the industry toward new identity standards: SAML for enterprise federation, OAuth and OpenID Connect for mobile and cloud, and security event tokens for asynchronous threats. Today, agentic AI is forcing many of the same questions again, only faster.

How should organizations authenticate and authorize autonomous agents? Should agents be treated as their own category of identity, as non-human identities, as workload identities, or as extensions of users? How do teams prevent agents from going rogue, leaving behind unmanaged access, or bypassing governance through side channels?

Joe, Nate, and James discuss SAML, OAuth 2.0, OIDC, token exchange, on-behalf-of flows, delegated access, step-up authentication, MCP gateways, agent governance, privileged access management, identity governance, and the identity fundamentals organizations need before AI adoption accelerates further.

The episode also looks at why CIAM is no longer only a login problem. Customer identity is a security problem, a business problem, a customer experience problem, and increasingly a boardroom conversation. As AI builders emerge across the business, including outside traditional IT departments, organizations need guardrails that help teams innovate without creating unmanaged risk.

You’ll hear practical perspectives on:

• Why AI is pushing identity standards back into a “Wild West” moment

• How existing standards like SAML, OAuth 2.0, OIDC, and token exchange still matter

• Why identity governance, privileged access, and access fundamentals are more important than ever

• How customer, workforce, partner, vendor, and non-human identity are beginning to overlap

• What organizations should ask system integrators and identity partners when planning for agentic AI

• Why identity platforms and open standards may help organizations stay adaptable as the AI identity landscape evolves

• Real-world use cases involving AI chatbots, portfolio recommendations, delegated tokens, trading tools, step-up authorization, and side-channel access risks

Whether you are responsible for CIAM strategy, identity architecture, application security, fraud prevention, customer onboarding, privacy, authentication, authorization, or AI governance, this episode is designed to help you think through what is happening now — not just what standards may emerge later.