Identity platforms now sit at the center of enterprise risk

At a glance

What this is: This is a position piece arguing that enterprise identity architecture must be rebuilt for humans, workloads, and agentic AI as a single control plane.

Why it matters: It matters because IAM, PAM, and identity architecture teams now have to govern machine-speed access, attribution, and policy enforcement without relying on assumptions built for older identity estates.

By the numbers:

• When AWS credentials are exposed publicly, attackers attempt access within an average of 17 minutes and as quickly as 9 minutes in some cases.

👉 Read Newcore's perspective on rebuilding identity architecture for humans, workloads, and agents

Context

Identity platforms now sit inside the attack path they are supposed to control. In a cloud-first, agentic environment, the key question is not whether identity is important, but whether the architecture can still distinguish and govern humans, service accounts, workload credentials, and autonomous agents without collapsing into shared trust.

The article argues that legacy IAM was designed for a different operating model and that bolting AI onto old federation, token, and review patterns does not solve the underlying governance gap. That is the core identity security problem: machine-speed access has outgrown human-paced control design.

For identity programmes, the issue is no longer just access administration. It is whether attribution, policy enforcement, and breach resistance can be maintained across a converged identity estate that now includes agentic AI and traditional non-human identities.

Key questions

Q: What breaks when governance relies only on quarterly access reviews?

A: Quarterly reviews miss the day-to-day drift that accumulates between certification cycles. By the time the review happens, the access graph may already have changed, so the programme validates yesterday’s state rather than today’s risk. That makes certification useful for assurance, but weak as a primary control.

Q: Why do shared secrets and replayable tokens increase enterprise risk?

A: They create portable trust that can be copied, replayed, or abused far beyond the original authentication event. In a converged identity estate, that means one stolen credential may unlock multiple systems, which turns credential compromise into a broad access problem rather than a single account problem.

Q: What do security teams get wrong about agentic identity?

A: The common mistake is assuming an AI client can be governed like a normal service account with one stable use case. In practice, agentic systems can change their tool choices and timing within a task, so static policy alone does not explain behaviour. Governance must cover the session, not just the credential.

Q: Who is accountable when identity infrastructure is the entry point?

A: Accountability usually spans platform owners, IAM operations, and security governance because the failure often sits in the trust layer rather than in a single endpoint. Frameworks such as NIST CSF and OWASP Non-Human Identity Top 10 help assign ownership for privileged access, federation integrity, and non-human credentials. The practical goal is clear responsibility for the systems that issue trust.

Technical breakdown

Why legacy IAM breaks under agentic identity

Legacy IAM assumes identities are relatively stable, access is reviewed on a human cadence, and policy can be enforced around known roles and request patterns. Agentic systems change that because they can act at machine speed, combine tools, and initiate access paths that were never explicit in the original architecture. The result is not just more traffic. It is a control mismatch between static provisioning logic and dynamic runtime behaviour. Identity providers, federation layers, and review workflows were built to answer who should have access, not what an autonomous system will decide to do next. Practical implication: treat agentic access as an architecture issue, not a feature request.

Practical implication: Map where current IAM logic assumes human-paced decisions and identify which controls fail when actions occur at runtime speed.

Why shared secrets and replayable tokens are brittle identity foundations

Shared secrets, bearer tokens, and weak assertion schemes concentrate trust in credentials that can be copied, replayed, or stolen. In modern enterprise environments, that is especially dangerous because workloads, integrations, and agents often authenticate without the friction or behavioural cues that help protect human sign-in flows. A single stolen signing key or exposed token can create enterprise-wide trust collapse if the architecture treats it as a valid proof of identity across too many systems. The article’s point is structural: identity systems should not depend on credentials that function as portable skeleton keys. Practical implication: reduce the blast radius of each authenticator and separate authentication proof from broad downstream trust.

Practical implication: Review where bearer-style credentials still unlock broad access and narrow their scope before they become enterprise-wide trust anchors.

What converged identity graphs change for humans, workloads, and agents

A converged identity model puts humans, workloads, and agents into one governance plane, but that only works if each actor type retains its own lifecycle, attribution, and policy logic. Treating agents like service accounts with a cosmetic label hides their runtime decision-making, while treating service accounts like people creates review noise and weakens accountability. The technical challenge is not simply unifying directories. It is preserving actor-specific controls while maintaining a shared policy and observation layer. That requires identity graphs, privilege boundaries, and telemetry designed to follow the actor type, not just the authentication event. Practical implication: build one control plane, but do not collapse the governance model into one generic identity class.

Practical implication: Design the identity graph so human, workload, and agent governance stays distinct even when policy is centrally enforced.

Threat narrative

Attacker objective: The attacker wants to turn identity infrastructure into a reusable trust broker that unlocks broader access than the original credential should ever have provided.

1. Entry occurs when attackers gain or abuse access through stolen credentials, replayable tokens, or compromised signing material that the identity stack treats as valid trust.
2. Escalation follows when one trusted identity, key, or assertion functions as a skeleton key across too many connected systems and policy domains.
3. Impact is enterprise-wide access abuse, because the identity platform itself becomes the shortest path from compromise to broad authentication trust.

Breaches seen in the wild

• Moltbook AI agent keys breach — Moltbook breach exposed 1.5M AI agent keys.
• Meta AI Instagram Account Takeover — 20,225 Instagram accounts hijacked via compromised Meta AI support chatbot with overprivileged access.

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 is no longer a feeder system for security tooling. It is the control plane the CISO has to defend directly. When attackers now enter through credential abuse, token theft, MFA bypass, or session hijack, the centre of gravity has moved from perimeter thinking to identity architecture. The implication is that identity governance cannot remain an administrative layer sitting underneath the real security stack.

Legacy access review assumptions collapse when the actor can move at machine speed. Quarterly review cadences were designed for access that persists long enough to be observed, certified, and revoked. That assumption fails when agentic systems can acquire, use, and release privileges within the same operational window. The implication is that review-based governance alone cannot describe or contain autonomous behaviour.

Converged identity requires separate governance logic for each actor type inside one control plane. Human identities, workloads, and agents can share policy infrastructure, but they do not share the same lifecycle, attribution model, or privilege dynamics. Treating them as a single category produces governance noise and blind spots at the same time. Practitioners should build for one fabric, but never one identity model.

Runtime trust debt is the right named concept for this problem. It is the accumulated risk created when identity platforms keep accepting assumptions that worked in pre-cloud environments, even though those assumptions no longer match how identities now behave. The more the estate depends on portable secrets, broad federation trust, and weakly bounded assertions, the more debt the architecture carries. Practitioners need to see this as structural liability, not incremental technical drift.

From our research:

• Two-thirds of enterprises have endured a successful cyberattack resulting from compromised non-human identities, with a quarter encountering multiple attacks, according to 2024 ESG Report: Managing Non-Human Identities.
• The average organisation believes more than 1 in 5 of their non-human identities are insufficiently secured, which shows how far governance gaps can spread before they are measured.
• As a next step, review 52 NHI Breaches Analysis for the recurring failure modes behind identity compromise patterns.

What this signals

Runtime trust debt: as more estates absorb agentic AI, the real risk is not just adding another identity type, but inheriting more assumptions that were never designed for runtime autonomy. Teams should expect review cadence, credential scope, and attribution models to become the first stress points in a converged identity programme.

With 72% of organisations already reporting or suspecting NHI breach exposure, the governance baseline is no longer theoretical. Identity leaders should use that pressure to separate actor types cleanly and align policy, telemetry, and lifecycle controls to the behaviour of each one.

For practitioners building toward Zero Trust Architecture, the practical change is simple: the identity layer is now the enforcement point that determines whether machine-speed access can be bounded before it turns into enterprise-wide trust collapse.

For practitioners

• Audit identity assumptions that still depend on human-paced review Identify where quarterly access reviews, manual certification, or ticket-based approval still define trust for workloads and agents. Replace those dependencies with controls that can evaluate access at runtime and limit how long any single credential remains broadly usable.
• Separate actor types in the identity graph Keep humans, service accounts, and agents in the same policy plane, but give each one distinct lifecycle rules, attribution, and privilege boundaries. This prevents one identity model from masking the governance differences that actually matter.
• Reduce reusable trust material across the stack Find shared secrets, replayable tokens, and signing keys that function like enterprise-wide skeleton keys. Narrow their scope, isolate where possible, and remove any authentication path that can unlock too many downstream systems from one compromise.
• Treat agentic access as an architecture decision Before adding more automation, define how agent actions will be attributed, bounded, and revoked when execution happens at machine speed. If the architecture cannot explain what an agent may do next, it is not ready for autonomous behaviour.

Key takeaways

• Identity architecture is now a direct security control surface, not just a provisioning layer, because attackers increasingly exploit the trust embedded in credentials and federation.
• The evidence shows that non-human identity compromise is already common, which makes legacy review and secret-handling patterns too slow for modern threat conditions.
• Teams should redesign governance around actor type, runtime behaviour, and trust boundary size, or the identity stack will keep carrying more risk than it removes.

Key terms

• Converged Identity Governance: A governance model that treats physical access and digital access as one coordinated assurance problem. It aligns ownership, lifecycle events, approvals, and reviews so that a person or contractor cannot retain one form of access after another has been removed.
• Runtime Identity Trust Debt: Runtime identity trust debt is the accumulation of hidden access paths that remain active after a workload, client, or integration changes. It appears when ephemeral design assumptions are not matched by offboarding, revocation, and lifecycle controls, leaving old trust in place.
• Identity Control Plane: An identity control plane is the governance layer that decides who or what can access systems and under what conditions. In practice, it coordinates authentication, authorization, privilege review, and lifecycle management across human and machine identities so access policy is enforced consistently across environments.

What's in the full article

Newcore's full article covers the operational detail this post intentionally leaves for the source:

• The vendor’s full framing of how a converged identity platform should handle humans, workloads, and agents in one control plane.
• The specific architectural claims behind hardware-bound, phishing-resistant, split-trust identity design.
• The migration argument for moving from legacy IAM to a rebuilt identity platform without rip-and-replace downtime.

👉 Newcore's full article expands on the converged identity platform model and the migration logic behind it.

Deepen your knowledge

NHI governance, agentic AI identity, and machine identity lifecycle are core topics in our NHI Foundation Level course, the industry's only accredited NHI security programme. If you are building or maturing an IAM or identity security programme, it is worth exploring.