Ethan LockshinSales at NVIDIA focused on AI adoption

The President of Auth0 predicts the Identity Perimeter over the next 3 years

Name: The President of Auth0 predicts the Identity Perimeter over the next 3 years
Uploaded: 2025-12-19T13:32:36+00:00
Duration: 1 min 35 s
Channel: Ethan Lockshin
Description: In a concise interview, Shiv, President and former CPO of Auth0, argues that AI security and identity security are inseparable and outlines three major…

19 Dec 2025 · 2 min

Watch on YouTube

1:35

In a concise interview, Shiv, President and former CPO of Auth0, argues that AI security and identity security are inseparable and outlines three major shifts that will shape the identity perimeter over the next three to five years. First, AI agents will function as new users, requiring the same authentication, authorization, and intelligence considerations as human users. Second, zero‑trust principles must be applied with fine‑grained, granular permissions to ensure agents only access data they are entitled to. Third, identity‑linked contextual memory will become portable, allowing an agent’s or user’s accumulated state to follow them across disparate tools and platforms. These trends suggest a future where identity governs not just human access but also the behavior and data continuity of AI agents.

Chapters

Shiv, you are the president and former Chief product officer of Auth0, the leader and pioneer in customer identity. Identity has become the new security perimeter. As cloud adoption, remote work and API architectures expand the threat surface. How do you see the identity perimeter evolving over the next three to five years?
So the headline is this, AI security is identity security. And so you cannot have one without the other. And so to avoid getting caught flat footed, you really need to prepare for three massive shifts. First, AI agents are kind of like the new users, right? So we need to think about agents just like we think about users, meaning they'll have access, they'll have permissions, and more importantly, they have intelligence. And so you really have to make sure that everything that you do in terms of authenticating and authorizing a user, you, you have to apply that to agents. The second thing is you still have to apply the principles of zero trust. And what that means is you really need fine grain authorization. And what I mean by that is like you really need to make sure you have granular permissions so that the agent only has access to information it is supposed to have. And then the last thing is, over time I think when I think of identity and when I think of context and agents, you can see today like you're starting to build memory in all of these products. And so at some point memory, I think will be tied to identity and that memory will need to be portable so that it can follow, so it can go with you across different tools that you're using.