TL;DR: AI is not inventing fresh attack paths so much as accelerating exploitation of legacy systems, unknown devices and weak OT foundations that organisations have tolerated for years, according to Forescout's analysis of Richard Horne's NCSC speech. The practical shift is from admiring visibility to fixing trust, resilience and access decisions before scale turns neglect into disruption.
At a glance
What this is: NHI Mgmt Group's analysis, drawing on Forescout's blog post: AI is accelerating the exploitation of legacy infrastructure, unknown devices and weak OT foundations, making resilience decisions more urgent than visibility alone.
Why it matters: Identity, access and operational resilience teams need to treat connected-device trust, access scope and lifecycle discipline as control problems, not as inventory problems.
Context
Cybersecurity programmes often fail because they treat visibility, patching and access as separate workstreams when the real problem is weak foundations across the whole connected environment. In OT-heavy settings, the gap is sharper because physical systems, supplier-managed devices and legacy infrastructure are frequently outside normal IT governance.
AI does not remove those governance gaps. It increases the speed at which attackers can find exposed systems, abuse over-permissive access and scale disruption across environments that were already fragile. That makes cloud security maturity a useful analogue for how practitioners should think about posture, even when the subject is OT and legacy infrastructure.
For most organisations, the starting position is atypical only in degree, not in kind: exposure is common, but the combination of discoverability, connectivity and operational dependence now raises the consequence of every missed control.
Key questions
Q: How should teams secure OT systems that share networks with IT?
A: Use network segmentation, strict allowlisting, and an accurate asset inventory to separate OT from enterprise IT and limit lateral movement. Place remote access through monitored jump hosts or industrial firewalls, restrict protocols to what each control system needs, and map dependencies before making changes. Shared services should have tightly scoped routes, not flat trust.
Q: Why do AI tools make legacy infrastructure more dangerous?
A: AI tools make legacy infrastructure more dangerous because they increase the speed, scale, and reach of actions taken against systems that already lack strong segmentation, patching, or monitoring. Attackers can use AI to find weak services faster, automate phishing and credential abuse, and adapt payloads to old protocols or exposed management interfaces.
Q: What is the best way to reduce risk from unmanaged connected devices?
A: The best way to reduce risk is to discover those devices continuously, assign ownership, and remove default or unnecessary network access. Use network access control, device profiling, and separate VLANs or firewalls so unmanaged IoT, shadow IT, and rogue endpoints cannot freely reach sensitive systems or credentials.
Q: What happens when convenience access is left in place too long?
A: Convenience access left in place too long becomes standing trust, which means broad, persistent access that is no longer tied to a current need. That increases exposure to privilege creep, forgotten service accounts, and reused credentials, especially when temporary exceptions, shared admin paths, or vendor access are never revoked.
Technical breakdown
Why weak foundations matter more once AI scales discovery
AI changes the economics of attack discovery. Rather than inventing new vulnerabilities, it helps adversaries enumerate exposed systems, correlate weak authentication paths and move faster through large environments with inconsistent hygiene. In practice, that means the attacker’s work shifts from manual reconnaissance to automated triage of obvious weaknesses. In OT and legacy estates, this is especially dangerous because long-lived systems often lack modern authentication, logging or safe update paths. The result is not a new class of flaw, but a much shorter time between exposure and abuse. Read Top 10 OT Security Risks.
Practical implication: Prioritise externally reachable legacy systems and unmonitored devices for remediation before using AI-assisted discovery against them.
How OT visibility differs from OT control
OT visibility tells you what is connected, which protocols are present and what assets exist on the network. OT control goes further by constraining who or what can talk to those assets, under what conditions, and with what blast radius if something goes wrong. That distinction matters because many industrial devices were never designed for modern trust boundaries. Once they are connected to enterprise networks, simple discovery is not enough. Without segmentation and explicit access decisions, the same asset inventory that improves awareness can also reveal a large attack surface. See the OT Security Platform Buyer's Guide.
Practical implication: Treat device discovery as the first step in containment design, not as evidence that the environment is understood.
Why convenience-based access becomes a systemic risk
Convenience access is access granted because it is easy to keep, not because it remains necessary. Over time, that pattern creates standing privilege, broad trust relationships and unclear ownership across humans, services and devices. In connected operational environments, those permissions can persist long after the business reason for them has disappeared. The technical problem is not only excess privilege, but also lack of lifecycle review, revocation discipline and context-aware authorization. Once AI accelerates exploitation, those stale permissions become a fast route from discovery to impact.
Practical implication: Inventory and re-justify every persistent access path that touches operational systems, especially where human convenience created the original exception.
Threat narrative
Attacker objective: The objective is to disrupt or control operational systems at scale by exploiting weak foundations faster than defenders can respond.
- Entry occurs through exposed, poorly governed connected assets that an attacker can discover quickly once AI-assisted scanning is applied to the environment.
- Escalation follows when weak access decisions or legacy trust relationships let the attacker move from visibility into control of operational systems.
- Impact is disruption of physical or operational processes, including the kind of shutdowns and service interruptions that affect refrigeration, utilities and manufacturing.
Breaches seen in the wild
- Schneider Electric Jira breach 2024: Credentials linked to a Lumma infostealer infection gave Hellcat access to Schneider Electric's Jira; 40GB and 400,000 user rows claimed.
- 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
AI governance debt: The real problem is not that AI creates entirely new flaws, but that it turns accumulated neglect into an operationally exploitable condition. Years of deferred upgrades, unknown assets and convenience-led access decisions now sit on top of one another, creating a governance debt that attackers can cash in quickly. Practitioners should treat AI as an accelerator of existing control failure, not as a reason to reframe the threat model from scratch.
OT is an identity problem as much as a network problem: Industrial environments now contain devices, controllers and service relationships that function like machine identities even when teams do not manage them that way. If those assets are not inventoried, authenticated and segmented as governed entities, they become trust anchors without ownership. That is why connected-device governance needs to include access scope, lifecycle control and containment, not just asset discovery.
Visibility without decision rights is insufficient: Knowing what is on the network does not tell you who should be able to reach it, what constitutes abnormal activity or how quickly a compromise should be contained. The speech points to a broader industry pattern where organisations know more than they control. Practitioners should close the gap between discovery and enforcement, or visibility will simply document the next failure.
Physical consequence changes prioritisation: The same weaknesses that would be tolerable in low-impact IT systems become much more urgent when they touch heating, refrigeration, water or manufacturing. That changes how boards should rank remediation, because the business impact is no longer limited to data theft or downtime. Identity, access and resilience teams should therefore elevate operational systems into the highest-risk tier when they share the same network fabric as enterprise IT.
Foundations must be built for the AI era, not retrofitted after it arrives: AI will accelerate both offence and defence, but it will not rescue architectures that were never designed for strong assurance. The organisations that fare best will be those that reduce unknown assets, remove convenience access and segment critical systems before attackers use AI to compress the time to impact. The priority is structural hardening, not reactive heroics.
What this signals
AI will compress exposure windows: As discovery and triage become faster, organisations will have less time to find and isolate weak OT assets before they are targeted. The practical consequence is that remediation queues must be built around exploitability and operational impact, not asset age alone.
Boards will increasingly need to ask whether their resilience strategy assumes visibility or control. In connected environments, those are different states, and only the latter meaningfully limits disruption when an attacker reaches a physical process.
For practitioners
- Harden exposed OT and legacy systems first Focus remediation on internet-reachable or internally discoverable systems that still lack modern controls, because those are the most likely to be found and abused quickly.
- Separate visibility from containment Use device discovery to drive segmentation, access restriction and exception removal, rather than treating inventory completeness as the end state.
- Remove convenience-based access paths Review persistent access into operational environments and revoke any permission that exists only because it was easier to keep than to redesign.
- Prioritise physical-process impact in risk scoring Rank heating, refrigeration, water and manufacturing controls above ordinary IT assets when they share trust boundaries or operational dependencies.
Key takeaways
- AI is best understood here as an accelerator of neglected infrastructure risk, not as a source of entirely new vulnerability classes.
- Operational technology becomes materially harder to defend once connectivity, weak access decisions and poor lifecycle discipline combine on the same network fabric.
- The most effective response is to move from discovery to containment, with segmentation, access reduction and resilience prioritisation for physical systems.
Related NHI Mgmt Group resources: OT Security Platform Buyer's Guide · Top 10 OT Security Risks · OT Security Maturity Model · Schneider Electric Jira breach 2024
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
MITRE ATT&CK and OWASP Non-Human Identity Top 10 address the attack and risk surface, while NIST CSF 2.0, CIS Controls v8 and NIST SP 800-53 Rev 5 set the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| NIST CSF 2.0 | ID.AM-01 — Physical Devices and Systems Inventoried | The speech centres on unknown devices and weak visibility across connected operational environments. |
| PR.AA-05 — Access Permissions, Entitlements and Authorizations | Convenience-based access and trust decisions are part of the risk described in the speech. | |
| Recommendation — Inventory connected OT and legacy assets so unknown devices cannot remain outside governance. Review and tighten access permissions to operational systems where convenience has replaced necessity. | ||
| CIS Controls v8 | CIS-1 — Enterprise Asset Inventory | The speech repeatedly points to devices nobody knows exist and the need to see what is connected. |
| CIS-6 — Access Control Management | The article ties operational risk to over-granted and long-lived access paths. | |
| Recommendation — Build and maintain an enterprise inventory of connected assets, including OT and IoT systems. Enforce access control management so only required connections and privileges remain in place. | ||
| NIST SP 800-53 Rev 5 | AC-6 — Least Privilege | The article explicitly references access granted for convenience rather than necessity. |
| Recommendation — Apply least privilege to connected systems and remove permissions that are no longer required. | ||
| MITRE ATT&CK | TA0007;TA0008;TA0040 — Discovery; Lateral Movement; Impact | The threat pattern is discovery of weak assets followed by movement into disruptive operational impact. |
| Recommendation — Map AI-assisted discovery and downstream disruption paths to ATT&CK and harden the exposed stages. | ||
| OWASP Non-Human Identity Top 10 | NHI-05 — Overprivileged NHI | Operational devices and connected services behave like governed machine identities when they retain excessive trust. |
| Recommendation — Reduce overprivileged machine and service access so connected assets cannot be abused at scale. | ||
Key terms
- Legacy Infrastructure: Legacy infrastructure is an existing environment built before current automation standards were in place. It may include manually configured resources, inconsistent naming, and limited code coverage. In practice, it creates migration and governance challenges because teams must import, validate, and standardise assets without disrupting live services.
- Operational Technology: Operational Technology is the hardware and software that monitors or controls physical processes such as manufacturing lines, utilities, and transportation systems. Unlike standard IT, OT prioritises uptime and safety, so identity controls must be precise enough to reduce risk without interrupting essential operations.
- Enterprise Asset Inventory: An enterprise asset inventory is a maintained record of hardware devices connected to the environment. It gives security teams visibility into what exists, what is authorized, and what needs to be removed or monitored. Without it, organizations cannot reliably enforce configuration, patching, or access controls across the full attack surface.
- Least Privilege: A security principle requiring that every identity, human or non-human, is granted only the minimum permissions necessary to perform its function. Least privilege is the single most effective control for reducing NHI blast radius.
Deepen your knowledge
NHI Foundation Level course, the industry's only accredited NHI security programme, covers NHI governance, machine identity security and secrets management. It helps security practitioners align identity controls with the operational systems and access paths that modern programmes depend on.
Published by the NHIMG editorial team on November 6, 2026.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org