Artificial intelligence is accelerating into a new phase. Beyond chatbots and automation, organisations are introducing agentic AI — autonomous systems capable of taking action, accessing data, and interacting with business systems much like human employees.
This shift brings significant efficiency, but also a major security concern:
Most AI agents currently operate with no formal identity, lifecycle management, or accountability.
During his visit to the Melbourne Grand Prix, Shiv Ramji, President of Auth0 at Okta, stressed the urgency of this oversight gap.
“As AI agents become the new workforce, with access to systems and data but no identity, it leaves businesses exposed.”
Why Traditional Security Models Don’t Fit AI Agents
Human users have predictable access patterns and well-defined roles.
AI agents, however, behave differently.
They may access:
- APIs
- Databases
- File systems
- Applications
- Even other agents
And they act non‑deterministically, adjusting their behaviour based on context and logic. This unpredictability challenges traditional role-based access models.
The Risks of Identity‑less AI Agents
Without structured identity governance, organisations face risks such as:
- Unmonitored or excessive access
- Data leakage
- Credential abuse
- Compliance failures
- Unauthorised actions at scale
Although AI agent adoption is growing quickly, governance frameworks are lagging behind.
What Effective AI Governance Looks Like
Okta recommends several foundational principles to close the gap:
- Assign identity to every AI agent
A unique, auditable identity with a defined function. - Strong authentication and access controls
Including just‑in‑time credentials and short‑lived tokens. - Fine‑grained, contextual permissions
Limiting access to only what is necessary. - Human‑in‑the‑loop oversight
Ensuring accountability remains with humans. - Scalability
AI agents can scale from 10 to 10,000 — identity systems must scale with them.
Okta is also contributing to emerging standards such as the MCP protocol to improve cross‑application access in AI environments.
Identity in Unexpected Places: The Grand Prix Example
Okta’s sponsorship of the McLaren team demonstrates the role of identity in delivering personalised experiences. Through Okta’s technology, fan interactions across apps, websites, and merchandise systems become more secure and seamless.
“Identity is a foundation of providing a 360‑degree fan experience.”
The principle holds true everywhere — whether for fans, employees, or AI agents.
CSB’s Perspective: Responsible AI Adoption Starts With Identity
At CSB, we see identity becoming the centrepiece of AI security. The risk isn’t AI’s capability — it’s uncontrolled access.
By grounding AI systems in identity, permissions, and human oversight, organisations can confidently embrace automation and innovation while maintaining control, safety, and transparency.