
Practical steps for defense operators to secure legacy systems, detect unauthorized activity, and preserve safe, reliable operations.
Galvanick Opinion.
November 9th, 2026
On October 8, 2026, the National Security Agency (NSA) released Hardening Department of War (DoW) Critical Operational Technology (OT) Environments with Zero Trust (ZT) Principles. The Cybersecurity Information Sheet (CSI) lays out practical steps to protect industrial systems against unauthorized access, lateral movement, and persistent adversary activity while preserving safe, reliable operations.
The CSI is primarily intended for National Security System (NSS), Department of War (DoW), and Defense Industrial Base (DIB) system owners and operators. NSA’s announcement also applies to technical leaders and experts, as well as IT and OT managers and administrators. The agency emphasizes that organizations can apply Zero Trust protections around legacy equipment without replacing their infrastructure.
Why NSA says to assume you’re already compromised
For years, OT security focused on keeping attackers out. NSA’s latest guidance says that’s no longer enough. The question isn’t whether an attacker could get in. It’s whether you would notice if they are already there.
Volt Typhoon is a Chinese state-sponsored group that breaks into U.S. critical infrastructure and stays there. Its intrusions typically start in IT networks and then pivot into OT. It rarely uses custom malware. Instead, it “lives off the land,” logs in with valid credentials, and uses the same built-in tools your administrators use. Then it waits.
The goal isn’t stealing data. It’s about gaining access, pre-positioning, conducting long-term reconnaissance, and then disrupting power, water, and other critical services later. An attacker this quiet can stay hidden for months. So NSA tells operators not to assume they’re clean. Assume someone is already inside and build defenses that can find them.
AI raises the stakes. NSA warns that attackers use it to speed up reconnaissance and exploit development, leaving defenders less time to respond.
This is why a successful login proves so little. An attacker with valid credentials and normal admin tools looks like an employee doing their job. Distinguishing them requires evidence about the user, the device, the activity, and whether it matches approved work.
What Zero Trust means in OT
Zero Trust starts from the same assumption: an attacker may already be inside. Being on the internal network earns no trust. Every access request must be checked against who is asking, what device they're using, which system they're accessing, and whether the behavior looks normal. It’s rechecked as conditions change.
OT makes this harder. Industrial equipment can run for decades. Much of it can’t support modern authentication, security agents, or regular patching. Restarting a device or interrupting its traffic can affect physical operations. That’s why NSA recommends compensating controls that protect legacy assets inside secure enclaves without modifying them.
The CSI draws on 84 target-level activities in DoW’s Zero Trust for Operational Technology Activities and Outcomes guidance. The CSI prioritizes 26 high-impact target-level activities from that set, chosen for their ability to provide immediate compensating control and operational visibility for legacy assets. These activities provide a starting point, but implementing them requires multiple controls, operational processes, and the resources to execute.
Key recommendations from the guidance
NSA’s CSI presents 10 high-impact OT Zero Trust mitigation steps that draw on 26 priority activities from the broader OT Zero Trust framework. Below, we summarize selected recommendations under five themes:
Control remote and privileged access. Remove direct OT connections to the public internet. Use phishing-resistant MFA (1.3.1), least privilege and role-based access (1.2.2), and privileged access management (1.4.1). Third-party and OEM access should be explicitly approved, time-bound, monitored, and terminated when the work ends (1.5.1).
Build inventories with operational meaning. Track devices (2.1.1), accounts including service accounts (1.1.1), applications (3.1.1), and configurations (2.6.2). An inventory should record each asset’s role, criticality, location, firmware, services, and communication relationships, so teams understand how systems support the process and what a security change might affect.
Protect communications as well as data. NSA treats OT data protection more broadly than encryption. Activity 5.4.3 covers availability, reliability, integrity, and authenticity of communications between OT assets. Where equipment cannot encrypt natively, gateways, brokers, or tunnels may help, subject to operational requirements.
Limit movement and unauthorized change. Segmentation should reflect asset function, criticality, and required communications (5.4.1), with approved connection paths and protocols enforced (2.2.1). Hardening includes changing default credentials, disabling unnecessary services, restricting who can modify controllers (1.7.1), and managing patches according to operational risk (2.5.1).
Monitor continuously and validate safely. NSA calls for endpoint detection (2.7.1), log parsing (7.1.2), an OT threat intelligence program (7.5.1, 7.5.2), file monitoring on engineering workstations (4.4.3), and hardened domain controllers. It strongly favors passive analysis through configuration reviews, log analysis, and SPAN/TAP monitoring, with any active testing confined to approved maintenance windows.
Where Galvanick contributes
Galvanick pulls endpoint, network, infrastructure, and application data into one behavioral detection engine. It runs in the cloud or on-premises. Collectors sit inside the environment they monitor and are built with operational resilience in mind:
Network sensors watch a copy of traffic through SPAN or TAP
Endpoint sensors run in user mode, with no kernel drivers or inbound connections
Analysis happens off the production hosts
Installs and updates follow the customer’s own change process
The table below lists a handful of activities from the CSI’s priority set, what each one requires, and how Galvanick supports it. This is only a sample and not a complete list of the platforms mapping:
Activity | Requirement | Galvanick’s Role |
Endpoint visibility (2.7.1) | Endpoint detection and response suited to OT | Records process and network activity on supported hosts so operators can detect and respond to threats |
Log analysis (7.1.2) | Logs detailed enough to model and prioritize common behaviors | Links data from several sources into deduplicated and contextualized findings with the supporting events attached |
Segmentation support (5.3.1 / 5.3.2, outcome 2) | Attribute and behavior-based enforcement checks for segmentation | Shows who and what is communicating so teams can decide what to allow and how to better inform their segmentation strategy |
File monitoring (4.4.3) | Monitoring of critical OT files with basic SIEM integration | Tracks file activity on engineering workstations and HMIs and sends findings to a SIEM |
Investigate activity that predates deployment
NSA’s assume-breach guidance raises a practical question: was the adversary already there? Galvanick can analyze available historical endpoint logs to surface evidence of activity that predates deployment. This helps teams investigate pre-positioning that monitoring only new events could miss.
Its detections target attacker techniques without requiring a baseline-learning period. As telemetry accumulates, teams also gain communication paths, asset relationships, and account activity to build and maintain the operational baselines NSA describes.
Establish whether activity matches approved work
Technical observations need operational interpretation. Not every alert is an attack. Galvanick checks flagged activity against your change-management records and can ask your operations team directly whether it was planned, so analysts stop chasing approved maintenance and focus on what’s real. An unexpected remote-access login or configuration change outside an approved maintenance window gives investigators a concrete event to validate.
This helps investigators judge whether the users, actions, assets, and timing they observe correspond to authorized work, without assuming that every unusual event is malicious or every familiar action is safe. It also supports the CSI’s emphasis on monitored third-party access (1.5.1) and enforced connection paths (2.2.1).
Where to start
Start by assessing your environment against NSA’s 10 mitigation areas and 26 priority activities, beginning with mission-critical systems. For each activity, ask whether your current telemetry could actually reveal unauthorized activity, including activity that may have started before you were watching.
Request a coverage assessment against the 26 priority activities, and explore Galvanick’s approach to OT Threat Detection.
References to NSA guidance describe how Galvanick capabilities relate to outcomes in a public CSI. They do not imply endorsement by NSA, the Department of War, or the U.S. Government.
Request a demo at Galvanick.com