Identity security, AI governance and third-party risk are converging

At a glance

What this is: This is SailPoint's conversation with KPMG on how identity security is shifting under cloud expansion, third-party exposure and AI-driven access demands.

Why it matters: It matters because IAM teams now have to govern human, NHI and autonomous access patterns in the same operating model without losing control of compliance or business continuity.

By the numbers:

• 60% of breaches are identity-based and a significant portion involve third parties.
• 72% of organisations have experienced or suspect they have experienced a breach of non-human identities.

👉 Read SailPoint's conversation with KPMG on identity security, AI governance and third-party risk

Context

Identity security now sits at the intersection of access governance, third-party exposure and operational resilience. In hybrid environments, the question is not simply who can log in, but which identities, vendors and machine actors can reach sensitive systems at the moment access is needed.

The article frames AI as a governance stress test rather than a standalone security problem. That matters for NHI, because autonomous agents and delegated third-party access both expand the number of identities that must be reviewed, scoped and monitored across the lifecycle.

Key questions

Q: How should security teams govern third-party access in hybrid identity environments?

A: They should treat third-party access as a lifecycle-managed identity domain, not a one-time onboarding event. Every vendor account, integration and delegated permission needs a named owner, a business purpose, periodic recertification and a clear offboarding trigger. The goal is to ensure access ends when the relationship or task ends, not when someone notices a problem.

Q: Why do autonomous agents complicate identity governance?

A: Autonomous agents complicate identity governance because they can make runtime access decisions and use permissions without a human operator pausing to review each step. That breaks assumptions built around stable users, predictable task windows and manual certification. Identity programmes need to know who owns the agent, what it may touch and when its access expires.

Q: What do organisations get wrong about right-time access?

A: They often treat right-time access as a provisioning problem instead of a governance state. If access is granted broadly and left in place, it becomes standing privilege with a longer exposure window. Effective right-time access depends on task scoping, expiry logic and a removal process that is actually enforced.

Q: What frameworks should teams use to align identity security with resilience?

A: Teams should map identity controls to the NIST Cybersecurity Framework 2.0 and use NIST-based control language to connect access decisions to protect, detect and recover outcomes. For NHI and delegated access, the most useful step is translating governance policy into provable ownership, scope and revocation behaviour.

Technical breakdown

Third-party identity governance in hybrid environments

Third-party risk becomes an identity problem when external users, vendors and integration accounts are allowed into the same operational fabric as internal staff. In cloud and platform-heavy environments, access is often distributed across SaaS apps, federated logins and service integrations, which makes ownership and offboarding harder to prove. The control failure is usually not lack of authentication, but lack of lifecycle discipline: who approved access, what it can reach, and when it should be removed.

Practical implication: map every third-party identity to a named owner, business purpose and removal trigger before access sprawl becomes unreviewable.

AI governance and autonomous agent access

The article's 'duality of AI' captures two different governance needs: using AI to improve identity operations, and governing AI agents that themselves need access. The second case is more difficult because an agent can act with delegated permissions across tools and data sets, creating a new class of non-human identity risk. If access is not constrained by task, time and scope, the governance model starts to resemble overprovisioned machine identity rather than controlled automation.

Practical implication: treat AI agents as governed identities, not just workloads, and require explicit scoping before they touch sensitive data.

Right-time access and overprovisioning control

The phrase 'right access at the right time' points to a core IAM tension: access is often granted for convenience long before the exact business task is known. That creates standing privilege, especially in complex environments where teams, vendors and integrations share systems. For identity programmes, the technical challenge is to reduce duration and breadth of access without breaking operations. That usually means aligning policy, certification and privileged access controls around actual task windows rather than broad role assumptions.

Practical implication: use time-bound, task-bound access rules to reduce standing privilege in vendor, workforce and machine identities.

Threat narrative

Attacker objective: The objective is to leverage trusted identities and delegated access paths to reach sensitive data and operational systems without triggering effective governance controls.

1. Entry occurs through identity-based access pathways in cloud, platform and third-party integrations, where the initial trust decision is often broader than the task requires.
2. Escalation happens when identities, including autonomous agents, are overprovisioned and retain access beyond the moment needed for work.
3. Impact follows when identity-based access is used to reach sensitive systems, creating compliance exposure, continuity risk and cross-domain blast radius.

Breaches seen in the wild

• LiteLLM PyPI package breach — LiteLLM PyPI supply chain attack, credentials stolen from users.
• Shai Hulud npm malware campaign — Shai Hulud campaign: npm malware exposed secrets on GitHub.

Read our 52 NHI Breaches Analysis report for a comprehensive view of breaches impacting Non-Human Identities including AI Agents.

NHI Mgmt Group analysis

Identity security is now a continuity control, not just an access control. The article is right to connect identity with business continuity and compliance, because the blast radius of a weak identity programme is now operational, not merely technical. When cloud adoption and platform integration increase the number of trusted paths into core systems, identity becomes the control plane for resilience. Practitioners should treat identity governance as part of service continuity planning, not as an isolated IAM exercise.

Third-party risk has become the weakest governance edge in hybrid identity estates. External identities often sit outside the normal employee lifecycle, yet they can hold the same or greater access than internal users. That breaks assumptions built into recertification, offboarding and ownership models, especially when vendor relationships change faster than access is removed. The implication is that third-party identity governance must be tracked as a distinct lifecycle domain, not as an audit afterthought.

AI access creates a new class of identity problem when agents operate like durable actors. Autonomous agents can consume sensitive data, call tools and propagate access decisions at machine speed, which means static role design is no longer enough. This is where the governance model starts to fail if it assumes access is assigned to a stable human operator behind the scenes. Practitioners need to reframe AI access as an identity lifecycle issue with its own approval, scope and retirement logic.

Right-time access is becoming the named concept that separates policy from practice. The article's emphasis on giving agents the right access at the right time points to a broader governance gap: many organisations can describe least privilege, but cannot enforce it against dynamic execution. That gap appears across workforce, third-party and machine identities alike. The practitioner takeaway is that access scope must be defined as an operational state, not as a one-time provisioning event.

Transformation expertise matters because identity failures are increasingly socio-technical. Technology alone does not resolve hybrid access sprawl, third-party entitlements or AI governance. Organisations need process redesign, ownership clarity and cross-functional control mapping to make identity decisions durable. The field should move away from tool-centric identity projects and toward governed operating models that can survive platform complexity.

From our research:

• Only 1.5 out of 10 organisations are highly confident in their ability to secure NHIs, according to The State of Non-Human Identity Security.
• Only 72% of organisations have experienced or suspect they have experienced a breach of non-human identities, according to The 2024 ESG Report: Managing Non-Human Identities.
• That confidence gap is why readers should also review Ultimate Guide to NHIs , Lifecycle Processes for Managing NHIs for lifecycle controls that make access reviews and offboarding enforceable.

What this signals

Right-time access is emerging as the more useful control concept than broad least privilege slogans. The operational question is whether access can be constrained to the exact business task without leaving durable permission behind. Organisations that cannot answer that question clearly should expect third-party access, workload identity and AI agent governance to converge into one review problem. For a broader control map, teams can anchor their programme to the NIST Cybersecurity Framework 2.0.

Third-party and autonomous identities are forcing IAM teams to think in lifecycle terms again. If an identity can outlive the task, the contract or the session, then governance must include owner, expiry and removal logic, not just authentication. The same pattern shows up in service accounts, partner access and AI agents that inherit data access. That is why the Ultimate Guide to NHIs , Lifecycle Processes for Managing NHIs remains relevant for programmes that are trying to operationalise revocation.

Only 72% of organisations have experienced or suspect they have experienced a breach of non-human identities, according to The 2024 ESG Report: Managing Non-Human Identities. That level of exposure means the problem is no longer awareness. It is execution discipline across permissions, lifecycle offboarding and delegated access, especially where vendors and machine actors overlap.

For practitioners

• Inventory third-party identities by business purpose Create a complete register of vendor, contractor and partner identities, including application scope, owning team and offboarding trigger. Reconcile it to the systems they can reach before the next access review.
• Define AI agent access as task-scoped lifecycle state Require a named owner, explicit approval path and retirement condition for each agent that can access sensitive data or tools. Do not rely on broad service roles to cover autonomous behaviour.
• Reduce standing privilege across hybrid platforms Review roles, API tokens and delegated accounts for access that persists beyond the actual task window. Replace durable access with time-bound entitlements where the business process allows it.
• Align identity controls to continuity and compliance evidence Tie access certifications, privileged access records and third-party reviews to evidence that can support audit and recovery planning. Use the NIST Cybersecurity Framework 2.0 to connect identity decisions to resilience outcomes.

Key takeaways

• Identity security is becoming a resilience issue because cloud, third-party and AI access now shape operational continuity.
• The biggest governance gap is not authentication, but whether organisations can prove who owns access, why it exists and when it should end.
• Programmes that cannot control right-time access across people, vendors and agents will keep accumulating standing privilege and audit exposure.

Key terms

• Third-party identity: A third-party identity is any account, token or delegated access path used by an external party such as a vendor, contractor or partner. These identities often sit outside workforce lifecycle controls, so ownership, expiry and offboarding must be explicit to avoid lingering access after the business need has ended.
• Right-time access: Right-time access means permissions exist only for the period, task and scope required to complete a specific job. In identity programmes, it is the practical control target behind least privilege, because durable access tends to become standing privilege unless revocation is automatic and enforced.
• Autonomous agent identity: An autonomous agent identity is the identity assigned to software that can decide what action to take, what tool to use and when to execute without human approval for each step. That behaviour changes governance because access can be created, used and discarded faster than human review cycles can respond.
• Standing privilege: Standing privilege is access that remains active beyond the immediate need for it. It increases blast radius because unused permissions can still be abused, especially in cloud, third-party and machine identity estates where revocation is delayed or ownership is unclear.

Deepen your knowledge

NHI governance, agentic AI identity, and machine identity security are core topics in our NHI Foundation Level course, the industry's only accredited NHI security programme. If you are responsible for identity security strategy or NHI governance in your organisation, it is worth exploring.

This post draws on content published by SailPoint: Identity security, AI governance and third-party risk. Read the original.