Processing...

GENuinely Secure Environments

A secure environment is not an office with a lock on the door and antivirus on the desktops. It is a physically and logically closed space in which the default answer to "can data leave here?" is no, and every exception is deliberate, documented and technically enforced. Most organisations describe themselves as secure whilst running the same laptops, the same public cloud services and the same consumer messaging apps as everybody else. A genuinely secure environment starts from the opposite position: nothing is permitted until there is a reason for it.

Secure Environment

Designed, Built and Lived In

GEN designs, builds and operates secure environments for clients whose work genuinely cannot tolerate leakage: financial institutions, defence and aerospace suppliers, legal and forensic practices, research organisations and businesses handling material whose disclosure would end a contract or a company. We also work inside them ourselves. Our own operational sites are run to the same standard, which is why we are able to describe the friction honestly rather than sell an idealised version of it.


The model rests on four principles, applied together. Applied separately they achieve very little.

  • Deny by default: people, devices, applications, protocols and destinations are all blocked until explicitly allowed.
  • No uncontrolled path in or out: every route for data, physical or electronic, is enumerated, and anything not enumerated is closed.
  • Own the whole stack: hardware, operating system, network, applications and communications, with no third party holding keys or configuration.
  • Assume the insider: controls are designed on the basis that an authorised person may attempt to remove data, whether deliberately, under duress or through carelessness.

Physical Security

Logical controls are irrelevant if somebody can walk into the room. Physical protection is layered, so that no single failure grants access to the working area.

  • Zoned access: the site is divided into public, reception, general and secure zones. Each boundary is a separate authorisation, and clearance for one zone confers nothing on the next.
  • Facial recognition and biometrics: entry to the secure zone is granted on biometric identity, typically facial recognition combined with a second biometric factor such as fingerprint or vein pattern. Biometric templates are held locally, never in a vendor cloud, and are stored as irreversible templates rather than images.
  • Airlocks and interlocked doors: single-occupancy interlocked vestibules prevent tailgating. One person is identified, one person passes, and the outer door is closed before the inner door releases.
  • Anti-passback: a credential that has entered cannot enter again until it has properly exited, which defeats credential sharing at the door.
  • CCTV and recording: full coverage of approaches, boundaries and circulation space, recorded to local storage inside the environment, retained under a defined policy and never streamed to a third party platform.
  • Intruder detection and monitoring: alarm systems to BS EN 50131 grading appropriate to the risk, with monitored response and tamper detection on every device and cabinet.
  • Visitors: pre-notified, identity checked, escorted at all times, issued a visibly distinct pass, and never left alone in a working area. Unescorted visiting does not exist.
  • Secure disposal: cross-cut shredding of paper on site, and physical destruction of failed or retired media with certificated evidence. Nothing legible or readable leaves the building.
  • Clear desk and clear screen: enforced, not encouraged. Nothing is left out, and nothing sensitive faces a window or a camera.

No Personal Devices, and the Right to Search

The single most effective exfiltration device ever built is the modern smartphone. It has a camera, a microphone, encrypted storage, an independent radio path to the outside world and a screen its owner is entitled to look at. No amount of network engineering compensates for allowing one into a secure area.

  • Personal devices are prohibited absolutely beyond the reception boundary. Phones, tablets, laptops, smart watches, fitness trackers, wireless earbuds, e-readers, cameras, voice recorders and removable media are all included.
  • Lockers at the boundary: personal items are deposited in individual lockers, RF-screened where the risk assessment requires it, before the biometric threshold.
  • Search on entry and exit: staff, contractors and visitors are subject to search of person, bags and outer clothing on the way in and on the way out, conducted at will and without prior notice.
  • Random and universal: searching is applied at random and applies equally to directors, engineers and cleaners. Exemption by seniority defeats the control entirely.
  • Consent up front: the search regime is a written condition of engagement, agreed before access is granted, so that it is never a matter for negotiation at the door. Searches are conducted respectfully, by a trained member of staff, and are logged.
  • No corporate mobiles either: a company-issued handset is exactly as capable as a personal one. The secure area has no mobile telephony of any kind.

The policy is unpopular for roughly a fortnight, after which it becomes unremarkable. It is also the control that clients ask about first, because it is the one they can see working.

Hardware Tokens, Held on Site

Every user is issued a hardware security token, and it is mandatory. No workstation, system or internal application can be logged into without one. There is no password only route, no fallback, no temporary exception and no emergency bypass, because a bypass that exists is a bypass that will be used.

  • The token never leaves the building: it is site property, held in the secure environment at all times, and it is not taken home, taken on leave or carried between sites.
  • Issued on arrival: security holds the tokens and issues each one to its named employee at the boundary, after biometric identification, as part of the same process that takes their personal devices into a locker.
  • Surrendered before departure: the token is handed back to security and checked in before the individual leaves the secure zone. A token not returned is treated as a security incident immediately, not at the end of the shift.
  • One token, one person: tokens are bound to an individual and are not pooled, shared or reassigned. Handing a token to a colleague is a disciplinary matter, and the issue and return log makes it visible.
  • Reconciled every day: the issue and return record is balanced against the physical access log at the end of each working day, so a token in the wrong place is noticed within hours rather than at the next audit.
  • Revocation is instant: a lost, unreturned or withdrawn token is deregistered centrally, and the accounts bound to it lose access at once regardless of whether the physical device is recovered.

The point of the arrangement is that the credential and the person are only ever brought together inside the secure environment, under supervision. A token held on site cannot be stolen from a bag on a train, cannot be used from a home network and cannot be quietly borrowed at the weekend. It is an authentication control and a physical custody control at the same time, which is what makes it worth the administrative effort.

The Network: Closed by Construction

The network is built so that the ordinary failure modes are not available. There is no browsing, no consumer cloud, no file sharing service and no general route to the Internet from a workstation.

  • Stateful firewalling throughout: not just at the perimeter. Every zone boundary and every inter-site link is policed by a stateful firewall with an explicit allow list, default deny inbound and, critically, default deny outbound.
  • No direct Internet access: workstations have no route to the public Internet. Where an external resource is genuinely required, it is fetched through an inspected, allow-listed proxy into a defined holding area, and the request is attributable.
  • Egress control is the real control: most environments spend their budget on keeping attackers out and leave the door open on the way back. We treat outbound traffic as the primary risk and log, restrict and alert on it accordingly.
  • Cloud messaging and transfer blocked completely: consumer and corporate messaging platforms, webmail, file sync services, code paste sites, screen sharing tools, remote support tools and personal storage are blocked by destination, by protocol and by DNS. Anything capable of moving a byte out is treated as a data exfiltration channel, because that is precisely what it is.
  • DNS is internal and sinkholed: resolution is provided by internal resolvers only. External resolvers, DNS over HTTPS and DNS over TLS to third parties are blocked, since an encrypted resolver is a tunnel with a respectable name.
  • Closed WireGuard mesh: sites, systems and authorised client endpoints are joined by a private WireGuard mesh with fixed keys. Peers are known to each other cryptographically, the mesh has no public entry point, and there is no route from the mesh to the open Internet.
  • Segmentation inside the mesh: membership of the mesh is not membership of everything on it. Traffic between peers is filtered on the same deny by default basis as traffic at the perimeter.
  • No wireless in the secure zone: wired connections only, with port security and physical control of patching. Where wireless is unavoidable elsewhere, it terminates outside the secure boundary.
  • Ports disabled: USB mass storage, optical media and other removable device classes are disabled at the operating system and, where the risk warrants, physically.

Linux Everywhere, Microsoft Nowhere

Every system in the environment runs Linux: servers, workstations, firewalls, hypervisors, telephony and appliances. There is no Microsoft software anywhere in the estate. No Windows, no Active Directory, no Exchange, no Microsoft 365, no Teams, no OneDrive, no Entra.


Microsoft is a threat in its very name. No other vendor has a comparable record: decades of critical vulnerabilities, a monthly patch cycle that has run for a quarter of a century without ever arriving at a secure product, and an architecture so porous that the fixes themselves routinely introduce the next set of holes. It has been described as trying to patch a fishing net with sugar cubes, and the description is fair. The problem is not that Microsoft is careless with individual bugs, it is that the underlying design is inherently weak, and no amount of patching changes what is underneath.


Then there is single sign-on, sold as convenience and delivered as a single point of catastrophic failure. The proposition is that one identity should unlock the mail, the files, the finance system, the customer records and the administrative console. When that identity is taken, and it is taken constantly, the attacker inherits the entire company in one step. There is no containment, no segmentation and nothing left to fall back on. Companies have lost their livelihoods to exactly this, and the architecture worked precisely as designed while it happened. A secure environment does the opposite: identity is compartmented, access is granted per system, and no single credential opens everything.


The platform is also built around telemetry, cloud identity and cloud storage, behaviours that are increasingly difficult to switch off and increasingly re-enabled by updates. A platform whose default posture is to send data to its vendor cannot be reconciled with an environment whose defining property is that data does not leave. Add a monoculture that attracts the overwhelming majority of commodity malware, licensing that assumes cloud identity, and update mechanisms that change behaviour without consent, and the case closes itself.


Linux gives us the properties the environment requires: source that can be inspected, builds we control, machines that do nothing we have not asked for, no dependency on a vendor's cloud for identity or licensing, and long-term stability without forced change. Systems are deployed from hardened, standardised images, run without local administrative rights for users, and are rebuilt rather than repaired.

A Curated and Limited Toolset

Users do not install software. There is no application store, no download, no "just this once". The environment carries a deliberately small catalogue of applications, each one assessed, packaged, version-pinned and allow-listed by name and by cryptographic signature.

  • Execution is permitted only for binaries on the allow list. Everything else, including anything a user manages to write to their own home directory, simply will not run.
  • Every tool admitted to the catalogue is assessed for what it can reach: does it phone home, does it sync, does it have a cloud tier, does it embed a browser, can it open a network socket to anywhere it chooses.
  • Interpreters, compilers, archivers, remote access tools and browsers are treated as privileged and are restricted to the roles that genuinely need them.
  • Updates are staged, tested and deployed from an internal repository. No machine reaches out to a vendor for a package.
  • A small catalogue is also a supportable one. Fewer applications means fewer versions, fewer interactions, fewer vulnerabilities to track and a helpdesk that knows every one of them properly.

Communication Inside the Environment

Communication is the hardest part to get right, because every convenient option is also an exfiltration route. The environment therefore provides a small number of internal channels and nothing else.

  • No telephones: no mobiles, and no external telephony from the secure zone. Where voice is required it is internal, over the mesh, using SIP-TLS and SRTP.
  • No email: not internally and not externally. Email is an unauthenticated, forwardable, attachment-bearing protocol designed in an era of mutual trust, and there is no version of it that belongs inside a closed environment.
  • Matrix for messaging: all internal messaging runs on a private Matrix (Synapse) homeserver inside the environment, with end to end encryption, federation disabled and no bridges to outside networks. Membership is provisioned centrally and rooms are scoped to projects.
  • Internal web systems for everything else: ticketing, documentation, case files, project records, approvals, timesheets and reporting are delivered as internal web applications, served over the mesh, with role based access and a full audit trail.
  • Attachments stay put: documents live in the internal systems and are referenced, not sent. There is no version of a file circulating in a mailbox, because there are no mailboxes.

Communication with the outside world happens deliberately, through defined and monitored gateways operated at the boundary, by named people, under a documented process. It is not something an individual workstation can do on a whim. Further detail on the transport layer is set out in GENuinely Secure Communications.

Getting Data In and Out

A closed environment still has to receive client material and deliver work product. The difference is that it happens through one known route rather than dozens of unknown ones.

  • A single transfer point: ingress and egress of data occur through a controlled transfer service at the boundary, never from a workstation.
  • Inspection and quarantine: incoming material is examined, held, converted where appropriate and released into the environment only once cleared. Active content does not cross the boundary.
  • Authorised release: outbound material requires named authorisation, is recorded against the job it belongs to, and is transferred over the mesh to the client rather than posted to a third party service.
  • No removable media: memory sticks, portable drives and optical media have no role in the process and are not permitted in the building.
  • Printing under control: printing is restricted, attributable, released at the device by the individual concerned, and logged. Printed material is either used within the zone or destroyed within it.

People, Vetting and Least Privilege

Technology addresses the outsider. Process addresses the insider, who is a far more realistic threat and, in the great majority of incidents, is not malicious but simply looking for an easier way to do their job.

  • Screening to BS 7858: identity, right to work, employment history, financial probity and criminal record checks, at a depth proportionate to the material handled, and repeated periodically rather than once at hire.
  • Least privilege by default: access is granted to the specific systems and case material a role requires, and to nothing else. Blanket access does not exist, and administrative rights are held separately from day to day accounts.
  • Joiners, movers and leavers: access is provisioned, changed and revoked as a single controlled process. A change of role removes the old access as well as granting the new.
  • Immediate revocation: departure removes credentials, biometric enrolment and mesh keys the same day, with the keys treated as compromised rather than reissued.
  • Confidentiality that outlives employment: obligations continue after the engagement ends, and staff are made aware that the material they handle belongs to clients who have contractual and statutory remedies.
  • Training that is specific: not a generic annual video, but instruction in the actual controls, the reasons for them and the correct route when the process is inconvenient.

Monitoring, Logging and Assurance

Controls that are not observed are assumptions. Everything of consequence is logged inside the environment, correlated, and reviewed by people who know what normal looks like.

  • Physical access events, biometric decisions, search records and visitor movements are logged and reconcilable against the electronic record.
  • Authentication, authorisation, privilege use, data access, transfer requests and print jobs are recorded centrally, on write-once storage held within the environment.
  • Behaviour is assessed by pattern rather than by single event: access at unusual hours, volumes out of keeping with a role, or activity from a location that does not fit the person concerned. Geofencing and pattern analysis at the network and service layer catch the things a credential check never will.
  • Backups are taken within the environment, encrypted, and held offline as well as online, so that recovery never depends on a third party holding the data or the keys.
  • Configuration is managed as code, reviewed before deployment, and drift is detected rather than discovered.
  • The environment is tested: penetration testing of the logical estate, and physical assessment including attempts at tailgating, social engineering and unescorted access.

Why it Matters: The Uncomfortable Truth

Almost every organisation that suffers a serious data loss had a security policy, a firewall and an antivirus subscription. What it did not have was a closed environment. Data left through a channel nobody had thought to enumerate: a personal cloud account used to work at home, a messaging app used because the approved route was slow, a photograph of a screen, a memory stick in a coat pocket.


Convenience is the attack surface. Every tool that makes information easier to move makes it easier to move out. The modern working environment has been built, deliberately and profitably, to maximise that ease, and the organisations that supply those tools are not the ones who carry the consequences when material escapes.


Not every organisation needs every element of this. A closed WireGuard mesh, biometric airlocks and searching people on their way out are proportionate to some businesses and absurd for others, and we would say so rather than sell it. But the principles apply to everybody: know what you hold, know every route by which it could leave, close the ones you do not need, and watch the ones you do. Every organisation can adopt those, and should, as far as its circumstances practically allow.


What does not work is the half-way house. You are either security aware or you are not. There is no partial position, because security is decided by the weakest route left open, not by the strongest one you have closed. An organisation with a hardened network and an open USB port has an open USB port. One with a strict device policy that quietly tolerates a manager using a personal cloud account has no device policy. Controls that are applied when convenient are not controls, they are preferences, and an attacker only has to find the day they were inconvenient. The decision to take security seriously is binary. What follows from it, how far you go and what you spend, is where proportion belongs.


A secure environment costs something real. It is slower to work in, it constrains how people communicate, and it removes tools they are used to at home. We say so plainly rather than pretending otherwise. What it buys is a defensible position: an environment where you can state, and demonstrate with evidence, exactly what routes existed for a piece of information to leave, and account for every one of them. Very few organisations can do that. The ones who genuinely need to are the ones we build these environments for.

Our Contractual Position

Section 5 of the GEN Framework Agreement sets out our confidentiality and data protection obligations, including the extended scope of confidentiality and the protection of personnel. The controls described here are how those obligations are met in practice rather than on paper. Where a client requires work to be carried out within a secure environment, that requirement is recorded in the Site Record under Section 2.6 and applies to every engagement on that site.


We do not present this as compliance with a certification scheme. The general purpose schemes an organisation is usually asked about are a long way below the standard described on this page, and passing one says very little about whether data can actually leave a building. Where a client carries genuine sector obligations, whether regulatory, contractual or imposed by their own customers, the environment is designed to those specific requirements, which are almost always stricter and considerably more concrete than any general standard.

Getting Started

A secure environment can be a single room, a floor, a building or a distributed estate joined by a private mesh. The scope follows the material being protected and the obligations attached to it, not the size of the organisation. We begin with an assessment of what actually needs protecting, what routes currently exist for it to leave, and what the business can realistically live with, then design and build accordingly.


If your organisation handles material where a leak would be terminal, contact our team to discuss the design, construction and operation of a genuinely secure environment.

Contact Us