Bangladesh Bank
2016- Sector
- Financial Services
- What was reported
- Attackers used the bank’s own payment messaging access to send fraudulent transfer instructions worth close to US$1 billion; most were blocked.
Control by Firevault / Cyber Control
Detection identifies activity. Incident responders still need an authorised, proportionate action. Control by Firevault can physically disconnect selected registered paths as part of a prepared response.
When an incident is active, which connection can your responders actually withdraw?
Incident
Path exposed
Outcome
Control taken
Approximately 120–150 ms
Management enable/disable operation
Around 2–3 seconds
SMS round-trip operation, good signal
March 2025 test observations, not guaranteed timings.
Why connection control matters
Waiting can leave a route available; acting too broadly can stop unaffected operations. The incident owner needs to understand both the isolation boundary and the services that depend on it.
A compromised segment may retain access to core systems while investigators assess the incident. Credentials, policy changes and session revocation remain important, but may not be the intended final enforcement step for a selected path.
Reported incidents in related sectors
These incidents show how a connection path can shape what happens next. They are not claims about what any single control would have prevented.
What Control changes
Control complements your network security, detection, identity management and incident response. It does not replace them. It gives their decisions a physical place to act, on the connections you select.
Control access
Supplier, engineering and partner routes are connected for approved work, to named systems, for a set window, and disconnected afterwards. Not open by default.
Control Plans
Control third-party connectionsControl connections in two approved stages
Control exposure
Each critical system keeps the connections it genuinely needs, inside a defined physical boundary. Paths it does not need are disconnected.
Control Plans
Enforce physical network boundariesProtect critical system connections
Control intervention
When your response plan calls for it, the selected path is physically disconnected on a decision by the people or rules you agreed in advance, and stays disconnected until an authorised release.
Control Plans
Contain ransomware spreadContain a live incidentAuthorise emergency disconnection
Where Control applies & Control modules
Responders investigate an affected user segment. Its registered uplinks can be withdrawn on the incident owner’s decision, while other segments continue only where the mapped dependencies permit.
Where do you need Control?
Fracture interrupts the selected connection, Execute carries out the authorised intervention, and Isolate defines the boundary in scope. Detection and remediation remain with existing security tools.
Concept illustration. Commanded, reported and independently verified states are shown separately because they are not the same thing, and what can be verified depends on your deployment. Disconnecting a selected path changes what can use that path. On its own, it does not contain an incident across a whole environment, and routes outside the plan, such as wireless or cellular links, need their own controls.
Control Blueprints
How Cyber Control applies in the industries where these connections matter most. Each Blueprint maps real systems, suppliers and operating constraints.
Transaction networks and trading floors.
Read the Financial Services Blueprint
Clinical networks and connected medical equipment.
Read the Healthcare Blueprint
Protect the cold chain. Control the connection.
Read the Retail & Supermarkets BlueprintFourteen Blueprints are published. Each deployment is designed around your own estate.
How Control works
Your tools detect, your people and policies decide, and Control changes the connection state of the selected path. Here is what that rests on.
Control changes the connection state of a selected path at Layer 1: connected, or physically disconnected. It does not inspect or alter the traffic on it.
Commands reach each control point over a configured command path, kept separate from the network being controlled. How each point behaves if that path is lost is agreed and tested.
Who can connect, disconnect or release each path is agreed in advance: named people, more than one approval where it matters, or rules and high-confidence alerts you set.
Commands, approvals and reported states can be recorded for your deployment, so you can show what changed, when, and on whose authority.
Before you deploy
Confirm the selected paths, supporting systems and failure behaviour in the actual environment. No generic drawing can substitute for this assessment.
Control governs the selected paths in your plan. It does not inspect traffic, and wireless, cellular and other routes outside the plan need their own controls.
No. They detect, filter and investigate. Control adds a physical way to act on the selected paths when your people, or rules you have agreed, decide to.
No. It changes whether a selected path is connected. It does not read or change what travels on it.
It can, which is why the dependencies, safe states and restoration steps for each path are agreed and tested with your operations and safety owners before anything goes live.
It controls selected connectivity. Disconnecting a path removes the connections an AI system was using and gives people an independent way to intervene. It does not necessarily stop a model or process running locally.
No. Control governs connections. Keeping data physically offline is Offline Secure Storage®, a separate Firevault product that can sit alongside Control.
Yes. Start with one connection: a supplier route, an IT–OT crossing, or the path to one critical system.
Your next step
Begin with one operational connection. Agree its purpose, intervention authority, dependencies and the checks that make restoration acceptable.
Discuss your Control requirements