As enterprises push AI agents into production, much of the conversation focuses on what these systems can do. But a more important question is emerging: Who is accountable for them?
That question sat at the center of "Intelligent Agent Identity: The Control Plane for Agent Sprawl and Emerging Privilege Risks," presented by Ashish Chandra of Wipro and a customer practitioner at SailPoint Navigate Austin. The discussion highlighted a challenge that extends well beyond agent security capabilities: many organizations are still struggling to establish ownership and accountability for the non-human identities they already have.
The session revealed a reality many enterprises are beginning to confront. AI-agent governance is often framed as a technology challenge, but the harder problem may be accountability. Discovering identities is becoming easier. Establishing ownership and governance at scale remains difficult.
The real challenge begins after discovery
For years, identity teams have focused on visibility. The assumption was that once organizations could discover their non-human identities, governance would naturally follow. In practice, discovery often exposes a different problem.
Many organizations can identify service accounts, machine identities, and application accounts across their environments, but they cannot consistently answer basic questions:
- Who owns this identity?
- Why does it exist?
- Is it still required?
- Who is responsible for reviewing its access?
Without clear answers, visibility creates awareness but not accountability.
One of the strongest themes from the discussion was that governance breaks down when ownership is unclear. Access reviews stall. Privileges accumulate. Lifecycle decisions become increasingly difficult to make with confidence.
In other words, discovery is not the finish line. It is where the harder work begins.
AI agents are exposing an existing problem
The industry often describes AI agents as a new governance challenge. In many ways, they are but unlike traditional machine identities, agents can make decisions, invoke tools, access systems, and act on behalf of users. That introduces new questions around authorization, oversight, and control.
However, the session highlighted a broader reality.
Many of the challenges organizations are now attributing to AI are rooted in longstanding identity gaps: unclear ownership, fragmented accountability, inconsistent standards, and unmanaged access.
AI agents are not creating these weaknesses; they are making them impossible to ignore.
Identity Governance alone is not enough
A recurring theme throughout the discussion was the difference between Identity governance and run-time authorization to know what an agent should be allowed to do.
As organizations deploy agents into environments with sensitive systems and data, runtime decisions become just as important as lifecycle governance. Many are moving toward models in which agents operate on behalf of users, with access decisions governed by policies, context, and existing entitlements rather than broad standing privileges.
This represents an important shift. Identity governance establishes accountability, authorization establishes control and enterprises need both.
Accountability is becoming the foundation of AI governance
The session introduced a practical framework built on two parallel responsibilities.
1. The first is governance: establishing ownership, managing lifecycle events, reviewing access, and monitoring risk.
2. The second is runtime enforcement: ensuring agents can perform only the actions they are authorized to execute.
Both are necessary but neither works effectively without the other. Most importantly, neither succeeds without accountability.
Technology can discover non-human identities. It can correlate information and enforce policy. But what it cannot do is answer who is ultimately responsible for a digital identity and the actions it takes. That remains a business decision.
The question leaders should be asking
Much of the AI conversation today centers on scale, productivity, and automation. Those conversations matter. But as agents gain access to critical systems and business processes, leaders should focus on a more foundational question:
Can we clearly establish who owns the agent, whose authority it is exercising, and what it is allowed to do?
The organizations that answer those questions consistently will be best positioned to scale AI with confidence.
The broader lesson from the discussion is that AI-agent governance may not be exposing a technology gap. It may be exposing an accountability gap.
Most organizations can discover non-human identities. Far fewer can consistently establish ownership, govern authority, and maintain accountability throughout an identity's lifecycle.
As agents become more capable and more autonomous, those foundations become increasingly important. The future of AI governance is not simply about making agents more capable. It ultimately depends on whether organizations can make them accountable.


