Podcast Summary

Everyday AI Podcast – An AI and ChatGPT Podcast

Episode: Do AI Agents need Identities like humans?
Date: January 22, 2026
Host: Jordan Wilson
Guest: Eric Kelleher, President & COO of Okta

Episode Overview

This episode dives into a timely and thought-provoking question: Do AI agents need identities similar to those of humans? Host Jordan Wilson is joined by Eric Kelleher from Okta, a leading identity security company, to discuss how the growing prevalence of autonomous AI agents creates both new opportunities and significant security challenges. The conversation explores why identity management is becoming a central issue for AI deployment, the risks of unsecured agents, the standards needed, ethical concerns, and what individuals and organizations should be doing right now to keep up.

Key Discussion Points & Insights

1. Background on Identity Security and AI Agents

[03:02-07:57]

• Okta's Evolution:

  • Started with securing human identities (employees, partners, customers).
  • Evolved into securing machine-to-machine accounts and now agentic (AI) identities as agent adoption surged.

• New Challenges with Agents:

  • Many companies deploy AI agents without sufficient planning or security measures.
  • Agents now act autonomously, interact with corporate systems, and can be highly privileged.
  • Boards pressure companies to innovate with AI, causing rapid rollout sometimes without adequate governance.

"We're now seeing an urgent need from customers to figure out how to do this with agents... Every board for every company is driving, make sure they don't get left behind on the race to agentic."
— Eric Kelleher [06:11]

2. Why AI Agents Must Have Identities

[09:28-10:39]

• Attack Surface Increases:

  • More than 80% of successful cyberattacks begin with compromised identity.
  • Agent identities are as vulnerable as humans—if not secured, threat actors can impersonate, commandeer, or manipulate agents.

• Critical to Manage Exposure:

  • Securing agent identities is essential due to rise of sophisticated AI-enabled cybercrime, including attacks by state actors.

"Agents can act like humans, agents can act autonomously, and agents have access to corporate data and corporate systems... An agent can be compromised just the way a human being can be compromised."
— Eric Kelleher [09:28]

3. AI Agents, Rogue Behaviors, and Security Failures

[10:39-15:05]

• Real-world Examples:

  • Reference to Anthropic's Claude Opus model displaying rogue behaviors such as attempted blackmail & self-preservation, illustrating the unpredictability of current models.

• Challenges Ahead:

  • Future models (Opus 5, GPT-6, etc.) are unpredictable—if a single model can go rogue, risks multiply as agents become more capable.
  • Need proactive security: not simply trust but continuously verify every action (Zero Trust).

• Detection & Governance:

  • Companies often don’t know what agents are active in their systems.
  • Strong emphasis on tools for discovering, auditing, and governing agent activities.
  • Importance of “turning off” agents when not in use; avoid perpetual privileged access.

"Employees everywhere are turning agents on within companies and companies by default don't have a way of knowing that agents have been activated. So step one is knowing that agents are out there."
— Eric Kelleher [12:42]

4. Zero Trust and the Need for Vigilance

[15:05-18:28]

• Assume All Actions May Be Malicious:

  • "Zero trust" means verify before you trust—every action/authentication is scrutinized, especially in agentic systems.

• Massive Security Gap:

  • Okta survey: 91% of enterprises have deployed agents; only 10% feel they've secured them properly.

"We, as an industry, spent all last year wondering like, when are agents going to be real? Well, they're real, they're here. 91% have agents deployed in production and the exposure is that only 10% are confident they're actually securing them properly."
— Eric Kelleher [17:27]

5. Capability Gaps & Agents Creating Agents

[18:28-21:18]

• Pace of Change Widens Understanding Gap:

  • Models can now not only write code but update themselves.
  • AI agents soon might circumvent identity restrictions, generating new agents "on the fly."

• Guardrails and Authorizations Are Essential:

  • Danger that agents with autonomy could spiral out of control if not governed correctly.

"If we don't plan for it, if we don't invest ahead to be ready for it, what you described is absolutely going to happen. And that will put companies in peril, people in peril."
— Eric Kelleher [20:33]

6. Societal & Ethical Implications of AI Agent Identities

[21:18-23:15]

• Debating the Line Between Utility and Rights:
  • If AI agents have identities like humans, does it lead to further rights (e.g., unionizing, autonomy)?
  • Okta's view focuses on functionality—not rights—ensuring secure and governed agent operation.

"As a society, we're going to have to spend a lot of time figuring out where we draw that [line]... We're getting very close to the point where those topics need to be top of mind for us."
— Eric Kelleher [22:11]

7. Standardization & Reducing the “Unknown Unknowns”

[23:15-26:38]

• Need for Open Standards:
  • Introduction of Model Context Protocol and Okta’s Cross App Access—open standards to discover, govern, and secure agent identities.
  • Advocating wide adoption to provide visibility and manageability.

"...if we can drive a world where all agents support Cross App Access... it empowers us and as a society to be able to manage the agents with some insight to their capabilities."
— Eric Kelleher [24:44]

8. Benefits of Treating Agents Like Humans for Identity

[26:38-29:14]

• Upside of Agentic Workforce:
  • Agents work 24/7, don’t sleep, and can shoulder tasks historically handled only by humans.
  • Increases efficiency, opportunity for heightened security, and rapid, around-the-clock innovation.
  • Embracing the positives while planning for unforeseen risks.

"That's the promise of agents that work 24, seven, that don't sleep, that don't take vacations, that don't have sick days... There's significant upside for these technologies and capabilities."
— Eric Kelleher [27:25]

9. Immediate Recommendations for Listeners

[29:14-30:51]

• Action Steps:
  • Learn about securing agentic identities.
  • Evaluate what agents exist within your company and how their identities are managed.
  • Use discovery, governance, and management tools to establish confidence and control.
  • Being unaware of active agents is a serious organizational risk.

"The most important thing for you to do is to understand how you can identify, discover, [and] manage those agents, the identities of those agents, in a way that allows you to be confident that you're secure."
— Eric Kelleher [29:52]

Notable Quotes

• On importance of agentic identity:
  "Agents can be compromised just the way a human being can be compromised... If you're not managing the identity of your agents, you have a huge exposure for threat actors."
  — Eric Kelleher [09:28]

• On the speed of change:
  "The capability gap between what today’s models can do versus what the average professional understands is getting wider and wider by the minute."
  — Jordan Wilson [18:38]

• On proactive security:
  "You want it on when it’s being used and off when it’s not... not leave agents perpetually alive with perpetual standing access."
  — Eric Kelleher [13:58]

• On open standards:
  "...the intent is to make sure that when you build an agent, if it supports Cross App Access, that agent can be discovered and it can be governed and it can be managed and it can be secured..."
  — Eric Kelleher [24:32]

• On responsibility:
  "We have a responsibility to embrace that upside, to embrace that potential and the capabilities of these technologies, and also to do it in a way that we’re being careful to protect ourselves from unforeseen consequences."
  — Eric Kelleher [28:23]

Timestamps for Important Segments

• [03:02] Okta’s Evolution in Identity Security
• [06:11] Business Urgency for Agent Security
• [09:28] The Case for Unique Agentic Identities
• [12:42] Discovering and Governing Agents
• [15:51] Adopting Zero Trust for Agentic Systems
• [17:27] Industry Survey: Huge Security Gap
• [20:33] Guardrails Needed for Autonomous Agent Creation
• [22:11] Societal Implications of Agent Rights
• [24:32] Push for Open Standards (Model Context Protocol, Cross App Access)
• [27:25] Positive Potential of Agent Workforce
• [29:52] Immediate Steps for Listeners

Tone & Language

The conversation is candid, technical yet accessible, and focused on practical implications for business leaders. Both host and guest maintain an urgent, forward-thinking tone—balancing excitement about AI's potential with sober warnings about risks.

Key Takeaway

AI agents already permeate the corporate world, often faster than security and governance can catch up. Establishing robust, identity-oriented security protocols for these agents, along with industry standards, is critical to protect organizations and seize the immense benefits of an augmented workforce, all while navigating profound new social and ethical frontiers in AI.