Q-Cyber Magazine
Identity and least privilege: The first pillar of enterprise zero trust
NIST and CISA place identity at the centre of zero trust: no implicit trust from location, and a per-request authorisation that should be no broader than the task requires.

Most serious enterprise incidents still involve a credential that could do more than it needed to do, for longer than it needed to do it, from a place the architecture treated as safe. Identity and least privilege are how that pattern is retired.
Privilege is a standing decision until it is made otherwise
In a traditional enterprise, a wide privilege is often granted once, at onboarding or at the creation of a service account, and then left in place. NIST's model asks for the opposite: a decision at request time, as narrow as the resource requires. CISA's identity pillar is the operational home of that decision. It is concerned with how users and services are identified, how strongly they are assured, how privileged access is governed, and how those facts remain visible.
Least privilege is therefore not only a user-account hygiene programme. It includes service-to-service credentials, standing administrator roles, emergency access, and the paths that skip ordinary controls. A "break-glass" account that never expires, or a shared operations login, is implicit trust by another name. So is a virtual private network that, once joined, can reach every administrative interface.
Compromised credentials remain a common way into an estate, which is why the ransomware guide groups prevention around initial access paths and why multi-factor and narrowed access appear so often in official advice. This article will not invent a success rate for any particular control. It will say that a privilege which is standing, shared or unexplained should be treated as a defect until it is otherwise.
Exhibit 1
Identity decisions a zero-trust programme should be able to show
| Question | Weak default | Target practice, drawn from published models |
|---|---|---|
| Who is requesting access? | A shared or reused credential | A unique subject identity, with a defined assurance level (CISA identity pillar; NIST 800-207) |
| From which device or workload? | Unknown or unmanaged | Device or workload posture is part of the decision (NIST: subject and device authorisation) |
| To which resource, for which action? | Broad access once 'inside' the network | Per-request, least-privilege authorisation to a named resource |
| For how long? | Standing privilege with no expiry | Time-bound or just-enough access, with governed exceptions |
| Can the decision be explained later? | Incomplete or unsynchronised logs | Visibility and governance, the cross-cutting capabilities in CISA's model |
Source: Qualitative target practice derived from NIST SP 800-207 (August 2020) and CISA Zero Trust Maturity Model 2.0. Not a maturity score for any named organisation.
Implications
- Inventory standing privileges, including service accounts and emergency access, and give each an expiry or a review date.
- Join identity to device or workload posture before granting a session to a sensitive resource.
- Replace "inside the VPN" with a resource policy for administration and for data.
- Keep decision logs that can explain who was allowed to do what, and for how long.
What this means for enterprise
Zero trust begins where privilege is still standing and unexplained. Enterprises that can show a current list of privileged identities, a per-request rule for their most important systems, and logs that make those decisions reviewable, are already doing the first and hardest part of the work.


