Zero Trust Architecture and ZTNA
Why the corporate network no longer protects access on its own, and what Zero Trust changes when you connect: identity, device, rights.
On this page
In 30 seconds
- Where it happens
- Every time you access a company application or data. Neither location on the network nor ownership of the device (company or personal) gives automatic trust[2].
- What the attacker wants
- To exploit a stolen identity, a phishing message or a hijacked session, attacks that do not depend on location on the network[1].
What it is
For a long time, company security looked like a castle: an internal network surrounded by a moat, where whatever sits inside is trusted. Microsoft describes these traditional models: they rely on network boundaries and assume that things inside the perimeter are safer than those outside[1]. They were effective against older threats such as network scanning, but they are no longer enough today, because modern attacks go through identity theft, phishing and session hijacking, without depending on location on the network[1].
Zero Trust responds to this finding. NIST describes it as a set of evolving paradigms that move defenses from static network perimeters to users, assets and resources[2]. It responds to remote users, personal devices (BYOD) and resources hosted in the cloud, outside a network owned by the company[2]. For employees, the change shows at every login.
Straight answers
What is Zero Trust, in one sentence?
Microsoft sums it up as "never trust, always verify": never trust, always check[1]. Access is granted only after verifying the person asking for it, the device used, its location and behavior, and its risk level[1]. This verification is not done once: it continues throughout the session[1].
What changes in practice for me?
The checks cover you, your device and the context of the connection[1]. Microsoft Entra conditional access policies analyze the user, the device and the location. They can require multifactor authentication (MFA) when security calls for it and stay discreet when it is not needed[3]. The device also counts: the more the organization knows about it (patch level, jailbreak or root status), the better it can justify blocking or allowing access[3]. From an unknown device, a restricted session may let you read emails or view files, without being able to download them to an untrusted device[3].
Does the VPN go away?
At Google, BeyondCorp began as an internal initiative to let every employee work from untrusted networks, without a VPN[4]. Its principles: access to services must not depend on the network you connect from, it is granted based on contextual factors tied to the user and their device, and it must be authenticated, authorized and encrypted[4]. The UK NCSC sets a similar rule: trust no network between the device and the service, including the local network[5]. What matters for employees therefore becomes their identity and their device, rather than being "inside" the network[4].
Do we need to buy a Zero Trust product?
Google presents BeyondCorp as "A mindset, not a toolset"[4]. The NCSC notes that many vendors and open source projects offer Zero Trust services, and that its principles help choose the combination of services that fits[5]. CISA adds that Zero Trust may require a change in an organization's philosophy and culture around cybersecurity[6].
How it works
Microsoft bases Zero Trust on three principles[1]. The first, verify explicitly, requires every access request to be authenticated and authorized using all available signals[1]. The second, least privilege, gives users and workloads only the access they need, for the shortest time[1]. The third, assume breach, designs controls expecting that attackers may already be active in the environment, to limit the impact of an intrusion and speed up detection[1].
NIST adds that authentication and authorization, of both the user and the device, are separate functions performed before a session to an enterprise resource is opened[2]. CISA takes up NIST's seven principles, including access granted session by session and a dynamic access policy[7]. At the NCSC, a policy engine evaluates each request with several signals, such as location and device state, and the user's identity and status[5].
Microsoft sorts these controls into pillars: identities, endpoints, data, applications, infrastructure, network and security operations[8]. For identities, every request starts with an identity check and the application of least privilege[8]. For endpoints, access depends on the device's health, compliance and risk[8]. NIST also specifies that applications and services have their own identities: control no longer rests only on IP addresses, subnets and the perimeter, but also on these identities[9].
What you do
- Respond to the MFA request. When a user appears on a new device or from a new location, responding to the MFA challenge is, according to Microsoft, one of the most direct ways to signal that this device or location is familiar to them[3].
- Switch to passwordless sign-in when it is offered. Microsoft cites FIDO 2.0 and phone-based passwordless sign-in, to reduce the risk of phishing and password attacks, especially for sensitive or privileged users[3].
- Use single sign-on (SSO). It avoids leaving copies of your credentials in several applications and helps avoid phishing and "MFA fatigue" caused by too many prompts[3].
- Know that the device weighs in the access decision. For Microsoft, the more the organization knows about a device, the more it can justify blocking or allowing access[3].
- Do not share your credentials. Microsoft lists credential sharing and poor security hygiene among the behaviors that expose the organization and that attackers exploit[1].
- Use the intended channel to request access. Microsoft describes requests, approvals and recertifications that leave a record of why each person has their rights[3].
- Think before granting consent to an application. Microsoft recommends restricting user consent and managing consent requests, to avoid needlessly exposing the organization's data[3].
Pitfalls
- Seeing it as a product to buy. Google speaks of a mindset rather than a tool[4]. Microsoft invites thinking in pillars rather than individual products or features[8].
- Leaving legacy exceptions. Older systems often rely on "implicit trust": access there is evaluated rarely, based on fixed attributes, which runs against adaptive trust evaluation[7]. The NCSC recommends identifying early the legacy services that do not support Zero Trust[5]. Microsoft asks to block legacy authentication, because stolen or replayed credentials are used there against protocols such as SMTP, which cannot respond to modern controls[3].
- Keeping two identity management systems. Microsoft states that this duplication reduces the available signals and lets malicious actors hide between the two engines[3].
- Neglecting privileged accounts. Microsoft specifies that privileged access is not limited to administration access: it includes application or developer access that can change how critical applications operate and process data[3]. It recommends Privileged Identity Management to secure these identities[3].
- Relying on complex passwords. Microsoft writes that traditional complex password policies do not prevent the most common attacks, such as password spray and replay after a leak[3].
- Assuming the usual protections carry over. A Zero Trust architecture changes how protections such as malicious site filtering and phishing protection are implemented, and these may come from other solutions[5].
Real cases
Google, BeyondCorp since 2011
Google dates the BeyondCorp mission from 2011 to today: to let every employee work from untrusted networks, without a VPN[4]. Google states that most of its employees use it every day for user- and device-based authentication and authorization, for core infrastructure and enterprise resources[4]. Google adds that BeyondCorp can now be enabled in almost any organization with Chrome Enterprise Premium[4].
US federal agencies, 2021 to 2024
Executive Order 14028, published in the Federal Register on May 17, 2021, requires US federal civilian agencies to develop a plan to implement a Zero Trust architecture[7]. OMB memorandum M-22-09, published in January 2022, asks agencies to reach cybersecurity goals by the end of fiscal year 2024[7]. On identity, these texts provide for centralized identity management, phishing-resistant MFA and at least one device-specific signal in authorization decisions[3]. CISA published version 2.0 of its maturity model in April 2023[7]. The model's overview page is now in CISA's archived content, which may no longer reflect current policy[6].
NIST, 19 example implementations, June 2025
In June 2025, NIST published the final version of its SP 1800-35 guide[10]. The NCCoE center worked on it with 24 collaborators to integrate commercial technologies and build 19 example implementations of a Zero Trust architecture[10]. The guide aims at authorized access to resources spread across on-premises sites and several clouds, for hybrid employees and partners who connect from anywhere, at any time and from any device[10].
Sources
- Microsoft, “Zero Trust as a security foundation”. https://learn.microsoft.com/en-us/security/zero-trust/zero-trust-overview (accessed 10 November 2026)
- NIST, “SP 800-207, Zero Trust Architecture”, 11 August 2020. https://csrc.nist.gov/pubs/sp/800/207/final (accessed 10 November 2026)
- Microsoft, “Securing identity with Zero Trust”. https://learn.microsoft.com/en-us/security/zero-trust/deploy/identity (accessed 10 November 2026)
- Google, “BeyondCorp Zero Trust Enterprise Security”. https://cloud.google.com/beyondcorp (accessed 10 November 2026)
- NCSC (UK), “Zero trust architecture design principles”. https://www.ncsc.gov.uk/collection/zero-trust/architecture-design-principles (accessed 10 November 2026)
- CISA, “Zero Trust Maturity Model”. https://www.cisa.gov/zero-trust-maturity-model (accessed 10 November 2026)
- CISA, “Zero Trust Maturity Model Version 2.0”. https://www.cisa.gov/sites/default/files/2023-04/zero_trust_maturity_model_v2_508.pdf (accessed 10 November 2026)
- Microsoft, “Overview - Technology pillars”. https://learn.microsoft.com/en-us/security/zero-trust/deploy/overview (accessed 10 November 2026)
- NIST, “SP 800-207A, A Zero Trust Architecture Model for Access Control in Cloud-Native Applications in Multi-Cloud Environments”, 13 September 2023. https://csrc.nist.gov/pubs/sp/800/207/a/final (accessed 10 November 2026)
- NIST, “Implementing a Zero Trust Architecture: High-Level Document”, 10 June 2025. https://csrc.nist.gov/pubs/sp/1800/35/final (accessed 10 November 2026)
Cite this article
Brain, “Zero Trust Architecture and ZTNA”, Cyberpedia, 2026. https:/
BibTeX
@misc{brain_cp227_en_2026,
author = {Brain},
title = {Zero Trust Architecture and ZTNA},
year = {2026},
howpublished = {Cyberpedia},
url = {https://startbrain.io/cyberpedia/en/articles/zero-trust-ztna-end-of-castle-and-moat/},
note = {CC BY 4.0}
}