Identity core design is failing human, NHI and agentic threat models

At a glance

What this is: This analysis argues that identity core design is now a primary attack surface, with help desk resets, session replay and unmanaged non-human identities exposing the limits of human-first controls.

Why it matters: IAM, PAM and NHI teams need to reframe identity from a login layer into a governed control plane that must survive vendor compromise, delegated access and machine-speed change.

Context

Identity core design is the control plane that decides who or what can authenticate, inherit trust and act across systems. The article argues that this core was built for a human-first world and now fails when attackers target help desks, session artifacts and non-human identities instead of login prompts.

The governance gap is not only technical. Organisations still rely on assumptions that access is stable, identities are known, and trust can be granted once then reviewed later. Those assumptions break down when identities include service accounts, API keys and AI agents that appear, act and disappear faster than human review cycles.

Newcore uses recent breach patterns to show that identity is no longer supporting infrastructure. It is the target, the blast radius and, in some cases, the point of compromise.

Key questions

Q: What breaks when help desk resets are treated as low-risk support tasks?

A: The reset path becomes a privileged access channel that attackers can target through social engineering. If proofing is weak, the help desk can hand over control more easily than the primary login system would, which turns recovery into the easiest way around the perimeter.

Q: Why do replayed session tokens create such high identity risk?

A: Because they represent an already authenticated state, not just a password equivalent. Anyone who obtains a valid token can impersonate that session until it expires or is revoked, so the real issue is protecting session artefacts, not only hardening initial login.

Q: How should security teams govern non-human identities that have persistent access?

A: Security teams should treat every non-human identity as a managed asset with an owner, an explicit purpose, a scoped privilege set, and a defined offboarding path. Persistent access should be replaced with time-bound or task-bound access wherever possible, and every credential should be traceable to the system or workflow it supports.

Q: What happens when identity platforms become the attack vector?

A: A compromise in the control plane can cascade across the customer environment if the platform holds too much trust. That is why architectures need separated authority domains, constrained recovery paths and clear limits on what the vendor can assert on a customer's behalf.

Technical breakdown

Why help desk resets bypass the identity perimeter

Help desk workflows often sit outside the strongest authentication path, but they still carry authority to reset credentials, recover accounts and restore access. That makes them a control plane in their own right. In the MGM Resorts case described in the article, the attacker did not break MFA or crack a password; they persuaded a support process to grant access. This is a classic trust-transfer failure. The system authenticated the wrong thing, then handed over the result. In identity architecture terms, the weak point was not cryptography but delegated trust without strong verification of the caller.

Practical implication: treat account recovery as a privileged workflow, not a convenience feature.

How replayed session tokens defeat authenticated access

A valid session token is proof that authentication already happened, which is why token replay is so dangerous. Once an attacker obtains a token, they do not need to log in again; they can reuse the authenticated state until the session expires or is revoked. The Okta support incident described in the article shows how diagnostic files and support tooling can become the source of that credential material. This is a governance failure around session lifecycle, support-data handling and trust propagation. The core issue is that many identity systems protect entry well but protect authenticated state poorly.

Practical implication: inventory where session material can surface and eliminate uncontrolled token exposure paths.

Why non-human identity governance now includes AI agents

Service accounts, API keys and machine credentials were already hard to inventory, and AI agents make the problem more dynamic. Unlike a static service account, an agent can act on behalf of a human, call tools, chain actions and be created quickly outside formal provisioning flows. That changes the identity model from a mostly known population to a fast-changing one with delegated authority. Governance based on annual review, manual discovery or static entitlements cannot keep up with identities that are ephemeral in lifespan but high-impact in privilege. The result is not just more identities, but less certainty about who is acting and under what authority.

Practical implication: represent agents as governed principals with explicit lifecycle, scope and revocation controls.

Threat narrative

Attacker objective: The attacker sought authenticated access that could be turned into privileged control and operational disruption without breaking primary login protections.

1. Entry occurred through social engineering of a help desk workflow, where the attacker posed as an employee and used a reset request to gain a foothold.
2. Credential access followed through support-side trust transfer and, in a separate incident described by the article, replayed session tokens extracted from diagnostic material.
3. Escalation happened when the attacker used the reset access or replayed session state to reach privileged systems and expand control across the environment.
4. Impact included operational disruption, with the MGM Resorts incident cited as taking computers offline for ten days.

Breaches seen in the wild

• Dropbox Sign breach 2024: A compromised back-end service account gave attackers Dropbox Sign customer data, including API keys, OAuth tokens and MFA information.
• Microsoft Midnight Blizzard breach: Midnight Blizzard (APT29) exploited legacy test account without MFA to breach Microsoft.

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 core design has become the control plane attackers want to reach first. The article shows that modern breaches do not need to break the front door when they can exploit the systems that create, restore or inherit trust. That is why identity governance now has to be designed around attack paths, not around the assumption that login is the main event. Practitioners should treat the identity core as a high-value security domain, not as plumbing.

Helpful support processes are now privileged workflows in disguise. The MGM Resorts case is not just a social-engineering story. It is evidence that recovery and reset paths can outrank the primary authentication stack in practical risk terms. In governance language, the control failed because the organisation trusted the reset channel too easily. Practitioners need to reclassify account recovery, help desk action and exception handling as privileged access surfaces.

Session trust is not the same as identity trust. The Okta support incident described in the article shows that a successfully authenticated session can still be abused if the token or artefact that represents it is exposed. That means the real boundary is not initial authentication but the lifecycle of authenticated state. Practitioners should assume that session artefacts, support bundles and diagnostic files can become control-plane inputs for attackers.

Ephemeral authority is a better named concept for the current risk than static privilege creep. The article's AI agent discussion shows why identity governance must move from reviewing long-lived accounts to governing identities that appear, act and disappear at machine speed. Traditional review cycles assume enough time exists to observe, certify and remediate access. That assumption fails when the identity exists only long enough to complete a task. Practitioners should redesign governance for identity lifecycles measured in sessions, not quarters.

Vendor trust is now part of identity design, not a separate procurement issue. The article argues that the identity platform itself can become the attack vector. That shifts the governance question from whether a platform is compliant to whether compromise of the provider can cascade into customer identity compromise. Practitioners should separate authentication authority, recovery authority and administrative authority so one compromise does not become total trust collapse.

What this signals

Control-plane design now matters more than login hardening. The article's core point is that attackers increasingly route around authentication rather than defeat it. That changes programme priorities for IAM, PAM and NHI teams: the deepest risk sits in the systems that issue, recover and replay identity state, not only in the sign-in flow.

Discovery is the difference between governance and guesswork. If teams cannot enumerate service accounts, API keys, machine credentials and agent identities, they cannot govern them. The practical implication is simple: identity programmes need continuous discovery across directories, cloud control planes and agentic workflows, or the hidden population will keep expanding outside policy.

Human-first processes are now an architectural liability. Recovery desks, support tooling and manual approval chains were designed for slower human operations. When applied to NHI and agentic identity lifecycles, they create windows where trust is granted faster than it can be reviewed. Practitioners should expect the next control failure to come from process design, not just from exposed credentials.

For practitioners

• Harden account recovery workflows Treat help desk resets, callback procedures and exception handling as privileged access paths. Require stronger proofing, dual approval for sensitive resets and explicit logging of who authorised each recovery action.
• Inventory every session-holding identity artefact Map where session tokens, support bundles, diagnostic files and exported logs can expose active identity state. Remove uncontrolled storage and restrict who can access troubleshooting outputs.
• Separate platform trust from customer trust Design the identity core so compromise of a vendor support system or control plane does not automatically expose downstream customer identities. Break single points of trust where the platform alone can assert authority.
• Govern non-human identities as first-class principals Give service accounts, API keys and AI agents explicit ownership, scope, lifecycle and offboarding rules. Discovery should be continuous, not an annual clean-up of accounts nobody can fully explain.
• Move review cadence to issuance time Use automated issuance, rotation and revocation for identities that may exist only briefly or act at machine speed. Manual review is too slow for identities that can create impact before the next governance cycle begins.

Key takeaways

• The article shows that identity failures now often come from trusted workflows, not from broken cryptography or obvious malware.
• Help desk resets, session token reuse and unmanaged non-human identities demonstrate how much attack surface lives inside identity operations.
• The strongest control response is to govern recovery, session state and non-human lifecycles as high-risk identity functions rather than background administration.

Key terms

• Identity Core: The identity core is the set of systems and controls that decide who is trusted and what they can access, including Active Directory, domain controllers, and associated authentication services. If this layer is compromised, attackers inherit the mechanisms that govern the rest of the environment.
• Refresh Token Replay: The repeated use of a long-lived OAuth refresh token to obtain new access tokens without asking the user to authenticate again. It turns a single captured credential into durable session access, which is why revocation speed and token telemetry matter so much in identity governance.
• Help desk privilege: The elevated access or authority support staff may need to complete identity recovery tasks. In practice, this becomes a security concern when the organisation uses privileged operators as a workaround for weak self-service coverage instead of keeping recovery tightly bounded and verified.
• Agent Identity: An agent identity is the set of attributes, credentials and permissions assigned to an autonomous software entity. It is treated as a non-human identity because it can authenticate, act on systems and accumulate access over time, which creates governance, audit and lifecycle obligations similar to other production identities.

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 building or maturing an IAM programme, it is worth exploring.