{"canonical_url":"https://engineer.company/glossary/zero-trust/","content_markdown":"# Zero trust\n\n## What it is\n\nZero trust is a security model in which nothing is trusted because of where it\nsits on the network. NIST SP 800-207, the US federal publication that set out\nthe model in 2020, defines it as a paradigm focused on protecting resources, on\nthe premise that trust is never granted implicitly and must be evaluated\ncontinually. Being inside the office network, on the VPN or in the same data\ncentre earns a request nothing on its own. Every request for a resource is\ndecided on its merits: who is asking, from which device, for what, and under\nwhat conditions. Written with a hyphen, zero-trust qualifies a noun, as in a\nzero-trust design, and the plan that puts the model to work is called a zero\ntrust architecture, or ZTA.\n\nNIST names the parties precisely. A subject is whoever or whatever makes a\nrequest: a user, a service or a device. A resource is anything worth protecting,\nand the publication counts printers and compute capacity alongside data. Between\nthe two stand three logical components. The policy engine decides whether a\nrequest is granted and logs the decision. The policy administrator carries it\nout, opening or closing the path between subject and resource and issuing the\ncredentials or tokens for that one session. The policy enforcement point sits in\nthe path itself: it lets the session through, watches it and ends it when told\nto.\n\nThe publication explains the shift with an airport. Passengers pass one\nscreening checkpoint, and beyond it travellers, crew and airport staff mill\nabout the terminal, all treated as trusted. That terminal is what NIST calls an\nimplicit trust zone: an area where everyone is trusted to the level of the last\ncheckpoint they passed. A flat office network works the same way. Zero trust\nkeeps the zone as small as possible by moving the decision and its enforcement\ncloser to the resource.\n\n## Why it matters\n\nThe older model draws one line around the organisation and trusts what is inside\nit. NIST's objection is plain: once attackers breach the perimeter, lateral\nmovement is unhindered, and unauthorised lateral movement has been one of the\nbiggest challenges for the agencies the publication was written for. The line\nhas also stopped matching the work. Staff sign in from home and hotel networks,\npersonal devices are allowed onto some systems, and much of what they use runs\nin someone else's cloud. NIST's assumptions follow from that: the private\nnetwork is no implicit trust zone, an attacker may already be on it, and remote\nstaff should treat the local network they connect from as hostile.\n\nWhat a team gains is containment. Under the third of NIST's seven tenets, access\nis granted per session with the least privilege the task needs, and\nauthenticating to one resource grants nothing on another. A stolen password or a\ncompromised laptop then reaches what that account and device are allowed to\nreach, at that moment and through one enforcement point, rather than the whole\ninternal network. The decisions also leave a trail. The policy engine records\nevery grant and every denial, so the question of who could reach a system, and\nwhy, is answered from records rather than from memory.\n\n## What it looks like in practice\n\nTake a purchasing system, the kind of application NIST suggests for a first\nmigration because a defined group of staff uses it, unlike email, which the\nwhole organisation depends on. A buyer opens it from a company laptop. The\nrequest reaches the enforcement point in front of the application, which asks\nthe policy engine. The engine's trust algorithm weighs the inputs NIST lists:\nthe request itself, including the device's operating system version and patch\nlevel; the subject database; the asset database; the resource's own\nrequirements, which may name an authentication assurance level or refuse\naddresses abroad; and threat intelligence. An illustrative decision record, with\nfield names of our own rather than any product's, might read:\n\n```text\nrequest   buyer-17 on laptop-0417 -\u003e purchasing app, read and write\nsubject   procurement team; signed in with a phishing-resistant key\nasset     enterprise-managed, OS patch level current, disk encrypted\nresource  needs AAL2 or higher, a managed device, a domestic source IP\nthreat    source address absent from the threat feeds consulted\ndecision  grant read and write for this session; log as approved\n```\n\nThe grant covers this one session and this one application. When the same\nbuyer, on the same laptop, opens the supplier contracts kept in a document\nstore, the engine makes a new access decision: authentication to the first\nresource does not automatically grant access to the second. Context counts\ntoo. One of NIST's examples is an accountant who normally works office hours\nand suddenly asks for the financial system in the middle of the night from a\nlocation the engine does not recognise: a contextual trust algorithm can raise\nan alert and require the subject to meet a stricter confidence level first.\nAnother is an HR employee who reads 20 to 30 personnel records on a normal day\nand then asks for more than 100.\n\nFor cloud services, NIST points out that clients can go to the service directly,\nwith the organisation's enforcement points checking policy on the way, rather\nthan being brought into the internal network over a VPN first. Every one of\nthese decisions depends on a clean picture of who should reach what, and NIST\nasks for a survey of assets, subjects, data flows and workflows before anything\nis built. That groundwork is the kind of consolidation described in our case on\n[access management across M365 and SSO](/portfolio/led-access-management-across-m365-and-sso/):\nidentities and access brought together across M365, 1Password, Red Hat SSO and\nOKTA SSO, least privilege enforced, and onboarding and offboarding standardised\nso that access is granted and revoked promptly.\n\n## Where it goes wrong\n\n**Buying it as a product.** NIST states that zero trust is a set of guiding\nprinciples rather than a single architecture, and that there is no single set of\ntools or services for a zero trust enterprise. A new remote-access gateway with\nthe label on the box changes little if every subject it admits still reaches the\nwhole network behind it. Fix: treat the move as, in NIST's words, \"a journey\nrather than a wholesale replacement\", one business process at a time. Judge each\nproduct against the tenets it is meant to satisfy, including whether its data\nformats would let you leave it for another provider later.\n\n**Writing policy before the inventory.** The policy engine can only decide from\nwhat it knows about subjects, assets and processes. Where that knowledge is\nincomplete, NIST warns, the usual result is a business process that stops\nbecause the engine denies requests for lack of information, and shadow IT, the\nsystems set up outside the IT department's management, makes it worse. Fix:\nsurvey and catalogue the assets, subjects, data flows and workflows first,\nshadow IT included, and only then write the criteria for a candidate process.\n\n**Administrators keep blanket access.** In many older designs, developers and\nsystem administrators can reach every resource, and NIST singles these accounts\nout for extra scrutiny. Moving everyone else behind per-session decisions while\nthe administrator accounts keep their standing reach leaves the most valuable\ncredentials as the shortest way in. Fix: send privileged accounts through the\nsame engine, give them the access their current task needs, and use logs and\naudits to learn their normal patterns so that departures from them stand out.\n\n**The policy engine becomes the new perimeter.** No resource talks to another\nunless the engine and the administrator approve it. Anyone with access to the\nengine's rules can make unapproved changes or plain mistakes, and an attacker\nwho disrupts the enforcement points or the administrator can stall the business.\nFix: log and audit every configuration change, run the enforcement in a properly\nsecured cloud environment or replicate it in several locations, and give the\nstored traffic data and policy the most restrictive access of all, from\ndedicated administrator accounts only.\n\n**Enforcing everything on day one.** Few policy sets are right on the first\npass, and NIST expects a tuning phase in which requests that should succeed are\ndenied. Switching straight to strict enforcement turns every gap in the policy\ninto staff locked out of work they need. Fix: run the new workflow in\nreporting-only mode for a while. Grant most requests, compare the logs with the\nintended policy, and enforce only the basics, such as denying requests that fail\nMFA or come from addresses known to be hostile. Tighten the policy once the\nnormal pattern is understood.\n\n**Treating MFA as the finish line.** MFA lowers the risk from a stolen password,\nbut NIST points out that an attacker holding valid credentials, or a malicious\ninsider, can still reach whatever the account is allowed to reach. Fix: keep\neach account's reach narrow, and use a contextual trust algorithm that weighs\ntime, location and volume, so that valid credentials used in an unusual way\nstill raise an alert.\n\n## Further reading\n\n- [NIST SP 800-207, Zero Trust Architecture](https://csrc.nist.gov/pubs/sp/800/207/final)\n  — published in August 2020: the definition, the seven tenets, the logical\n  components, the threats and a road map for migrating.\n- [NIST SP 1800-35, Implementing a Zero Trust Architecture](https://csrc.nist.gov/pubs/sp/1800/35/final)\n  — the NIST practice guide published in June 2025, describing 19 example\n  implementations built with 24 collaborators.\n- [NIST SP 800-207A](https://csrc.nist.gov/pubs/sp/800/207/a/final) — published\n  in September 2023: access control for cloud-native applications across\n  several clouds, based on API gateways, sidecar proxies and service identities\n  rather than network addresses.\n- [NIST SP 800-63B-4](https://pages.nist.gov/800-63-4/sp800-63b.html) — the\n  authentication assurance levels a resource's requirements can name, of which\n  AAL3 requires phishing resistance.\n\n\u003chttps://engineer.company/glossary/zero-trust/\u003e\n","date_modified":"2026-11-09T21:50:25Z","description":"How zero trust swaps trust in the network for a decision on every request, what NIST's policy engine weighs, and where migrations stall.","formats":{"gemtext":"https://engineer.company/glossary/zero-trust/index.gmi","gopher":"https://engineer.company/glossary/zero-trust/gophermap","html":"https://engineer.company/glossary/zero-trust/","json":"https://engineer.company/glossary/zero-trust/index.json","markdown":"https://engineer.company/glossary/zero-trust/index.md","pdf":"https://engineer.company/glossary/zero-trust/index.pdf"},"kind":"page","language":"en","title":"What is zero trust?","url":"https://engineer.company/glossary/zero-trust/","version":1}