Identity and integrity risks are expanding across OT and AI workflows

At a glance

What this is: This is a threat landscape analysis showing that adversaries are increasingly targeting identity, operational coordination, engineering data, and AI-connected workflow layers rather than only traditional endpoints or control systems.

Why it matters: It matters because IAM, PAM, NHI, and OT security teams now need to govern access paths across contractors, workloads, integrations, and intermediary services, not just logins and network zones.

Context

Industrial and OT risk is no longer confined to direct device compromise. When engineering tools, simulation outputs, procurement systems, and AI-integrated workflows shape real-world operations, integrity failures in upstream systems can create downstream operational impact. That makes identity governance and access control part of resilience, not just IT hygiene.

The common thread in this analysis is the collapse of trust boundaries across operational systems. Once identities, shared credentials, contractor access, and intermediary services can influence critical workflows, traditional perimeter controls lose explanatory power and control has to move closer to issuance, authorization, and transaction time.

Key questions

Q: What breaks when contractors, workloads, or integrations can reach OT systems through trusted paths?

A: The control boundary breaks. Once a contractor account, shared credential, or integrated service can reach operational systems without tight scoping, the environment stops distinguishing intended action from compromised action. Attackers can move from initial access to manipulation inside trusted workflows, which makes containment depend on identity controls, not just network segmentation.

Q: Why does identity risk increase when AI tools sit inside operational workflows?

A: Because the tool can become a privileged intermediary. If an AI integration can access prompts, credentials, APIs, or connected systems, compromise of that intermediary can expose more than a single application. The governance problem is transitive trust, where one credentialed layer can amplify access across many connected services.

Q: How can organisations tell whether OT access controls are actually working?

A: Look for evidence that access is issued only on demand, expires automatically, and can be tied to a named user, task, and session record. If audits still require manual reconstruction, the control is not working at the governance level. Strong OT access management produces verifiable traces, not just fewer help desk tickets.

Q: What should organisations prioritise first in OT and AI-connected environments: segmentation or identity governance?

A: Identity governance comes first when trusted access paths already exist. Segmentation helps, but it does not remove risk from shared credentials, delegated access, or intermediary services that are already inside the workflow. Practitioners should start with scoped permissions, revocation, and continuous authorization, then layer network controls around that model.

Technical breakdown

Why upstream engineering data is now part of the attack surface

Industrial sabotage does not need to hit a PLC first. If an attacker can alter simulation outputs, engineering calculations, or design data, they can bias operational decisions before any direct disruption occurs. That makes data integrity a control problem, not just a storage problem. In these environments, trusted users and processes often have broad write access to systems that shape production logic. Once that trust is abused, the damage can remain hidden because the output still looks technically valid while the underlying assumptions have been corrupted.

Practical implication: Treat engineering and simulation platforms as governed control surfaces, with strict authorization and traceability on who or what can change outputs.

How trusted integrations expand identity risk in OT and AI workflows

Modern OT and AI environments rely on chained trust between users, contractors, services, APIs, and intermediary platforms. OAuth grants, workload identities, and shared service access can let one compromise propagate across otherwise separate systems. That is why the problem is not just authentication at the front door, but trust continuity across every hop. In multi-hop workflows, a compromised intermediary can observe, alter, or replay interactions while still appearing legitimate to adjacent systems. The result is a larger and more distributed identity plane than most governance models assume. Read the Ultimate Guide to NHIs: What are Non-Human Identities.

Practical implication: Model contractor access, OAuth permissions, and workload identities as transitive trust relationships that need explicit scoping and review.

Why operational resilience now depends on identity enforcement at the control layer

Identity-based enforcement matters in OT because many environments cannot patch, segment, or take systems offline quickly enough to rely on prevention alone. When exposure windows are long, the practical defense is to reduce what an authenticated user, workload, or service can do after access is granted. This is especially important where remote engineering, vendor support, or AI-assisted tooling can reach critical processes. Zero Trust in this context is less about network location and more about continuously verifying every interaction against policy, context, and role. See the NHI Lifecycle Management Guide.

Practical implication: Apply identity-aware authorization to limit post-authentication actions across engineering, orchestration, and remote access pathways.

Threat narrative

Attacker objective: The attacker aims to gain durable influence over operational systems by abusing trusted access paths, manipulating control logic, or corrupting the integrity of critical workflows.

1. Entry occurs through trusted operational pathways such as contractor access, shared credentials, compromised integrations, or upstream engineering tools rather than through obvious perimeter intrusion.
2. Credential or trust abuse lets the attacker operate inside legitimate workflows, where access to simulation data, PLC interfaces, procurement systems, or AI intermediaries can be exercised as if authorised.
3. Escalation happens when the attacker uses that legitimate position to manipulate outputs, persist in engineering environments, or influence downstream operational decisions.
4. Impact is achieved through disrupted operations, degraded integrity, hidden manipulation, or widened blast radius across OT, supply chain, or AI-integrated systems.

Breaches seen in the wild

• tj-actions/changed-files compromise 2025: A stolen bot token let attackers poison tj-actions/changed-files so pipelines printed their CI/CD secrets to public logs (CVE-2025-30066).
• CISA Private-CISA GitHub leak 2026: A CISA contractor's public GitHub repo exposed AWS GovCloud admin keys, Artifactory credentials and plaintext passwords for six months.

Read and download The State of NHI & AI Agent Breach Report 2026, covering 200+ breaches impacting Non-Human Identities including AI Agents.

NHI Mgmt Group analysis

Industrial integrity is now as important as industrial availability. The analysis shows attackers focusing on engineering data, simulation outputs, and operational coordination systems because those layers shape decisions before a control system is touched. That changes the governance problem from preventing obvious disruption to protecting the logic that makes operational decisions trustworthy. Practitioners should treat upstream data integrity as a first-class security objective.

Identity has become the control plane for distributed operational risk. The article repeatedly points to shared credentials, contractor access, trusted integrations, and intermediary services as the real paths into critical environments. That is exactly where traditional perimeter models break down, because trust now moves with the identity rather than with the network boundary. NHI lifecycle management becomes relevant wherever service, contractor, and workload access persists beyond the moment it was intended to be used.

AI is acting as a force multiplier for both attack speed and attack reach. The risk is not just that AI helps attackers find vulnerabilities faster, but that AI-integrated workflows create new intermediary trust layers that can be compromised or misused. That expands the attack surface into APIs, OAuth permissions, and proxy-style services that sit between systems. Security teams should assume that every added orchestration hop is also an added governance obligation.

Zero Trust in OT has to be evaluated at the point of action, not at the point of login. The article’s central implication is that identity checks at the edge are insufficient when attackers can remain inside trusted environments and act through legitimate pathways. The real question is whether systems continuously constrain what an identity can do after authentication. Practitioners should prioritise transaction-time enforcement over static trust decisions.

Multi-hop trust has become a named governance gap. Hidden intermediaries in AI and operational stacks create a transitive trust model where one compromised component can observe or alter downstream interactions without obvious failure. That is a specific governance failure mode, not a generic cyber risk. Teams should map these hidden trust layers explicitly and treat them as part of identity governance, not as invisible plumbing.

What this signals

Multi-hop trust is the new hidden attack surface. As AI tools, vendor services, and operational platforms stack more intermediary layers into a single workflow, security teams need to map who can see, alter, or relay each interaction. Without that map, identity controls remain partial and the real trust boundary stays invisible.

The practical shift for critical infrastructure is toward control at the moment of action. That means policy has to follow the identity through engineering tools, orchestration layers, and external integrations rather than stopping at authentication. The programme implication is straightforward: if you cannot explain each trust hop, you cannot govern the workflow.

OT and AI convergence will keep exposing gaps between theoretical access models and actual operational pathways. Teams that still think in terms of static zones will miss the more important question of whether an identity can influence a process after it is already trusted.

For practitioners

• Govern engineering outputs as protected assets Restrict who can modify simulation data, calculations, and design artefacts, and keep immutable audit trails for changes that affect operational decisions.
• Inventory contractor and shared access paths Map every contractor device, shared credential, remote engineering route, and vendor integration that can reach OT or logistics systems, then eliminate unmanaged persistence.
• Constrain OAuth and intermediary service permissions Review AI tool integrations, delegated permissions, and proxy layers so a compromise in one service cannot automatically expose credentials or broader workflow access.
• Apply continuous authorization to critical workflows Require policy checks at action time for PLC access, procurement actions, and orchestration requests instead of assuming a prior login is still trustworthy.
• Test blast-radius assumptions across OT and AI layers Use scenario testing to prove whether a compromised integration, workload, or contractor account can reach beyond its intended scope.

Key takeaways

• The core risk is no longer just disruption. Attackers are increasingly targeting the integrity of engineering data, control logic, and AI-connected workflows that shape operational decisions.
• Trusted integrations, contractor pathways, and intermediary services widen the attack surface because they let a compromise propagate through legitimate access paths.
• The strongest defensive lever is identity-centric enforcement at the point of action, backed by scoping, revocation, and traceable control over who or what can influence critical outputs.

Related NHI Mgmt Group resources: NHI Lifecycle Management Guide · Ultimate Guide to NHIs: What are Non-Human Identities · Top 10 NHI Issues

Key terms

• Operational Trust Boundary: The set of systems and processes surrounding a transaction that can influence whether a legitimate action becomes a malicious one. In crypto, this includes DNS, front ends, build pipelines, admin panels, and RPC endpoints, all of which can redirect or alter authorised activity.
• Transitive trust: The hidden risk created when one trusted app inherits confidence from another trusted relationship. In SaaS environments, approving a third-party tool means trusting its hosting, storage, developers, and connected services, which widens the attack surface beyond the original login event.
• Identity-Aware Enforcement: Identity-aware enforcement is policy decisioning that uses who the user is and which account or context they are using. For browser AI, it helps distinguish sanctioned corporate use from unmanaged personal sessions and applies different controls based on that distinction.
• Engineering Data Integrity: Engineering data integrity is the assurance that calculations, models, simulation outputs, and design artefacts remain accurate and unaltered by unauthorised actors. In industrial settings, compromising this layer can influence operational decisions without causing immediate outages, which makes the risk harder to detect and more damaging over time.

Deepen your knowledge

NHI Foundation Level course, the industry's only accredited NHI security programme, covers NHI governance, machine identity security, and workload identity in a way that supports both infrastructure and security teams. It helps practitioners connect identity controls to the operational systems their programmes depend on.