Traditional security was built around a boundary: a firewall at the edge, antivirus on the devices inside, and trust for anything on the internal network. Those controls still matter. The problem is the assumption underneath them.
The perimeter assumption no longer holds
NIST describes zero trust as “an evolving set of cybersecurity paradigms that move defenses from static, network-based perimeters to focus on users, assets, and resources.” Its core idea is that no access is trusted simply because of where a request comes from or who owns the device. NIST SP 800-207: Zero Trust Architecture
People now work from anywhere, core systems run in cloud services, and suppliers connect directly into business processes. Much of what needs protecting no longer sits behind the firewall at all.
How breaches actually begin
Verizon’s 2025 Data Breach Investigations Report, which analysed 12,195 confirmed breaches, found that “credential abuse (22%) and exploitation of vulnerabilities (20%) continue to be the leading initial attack vectors”. Exploitation of vulnerabilities as a starting point rose by 34%, and the share of breaches involving third parties doubled. Verizon 2025 DBIR announcement
A stolen password does not trip a firewall; it looks like a legitimate login. An unpatched internet-facing system is inside the boundary by design. A supplier’s remote access is trusted because it was approved. Each is a gap that perimeter controls were never designed to close.
The window is also narrowing. The UK National Cyber Security Centre assesses that “the time between disclosure and exploitation has shrunk to days”. NCSC: Impact of AI on cyber threat from now to 2027
What a 2026 security model adds
- Identity at the centre: strong, phishing-resistant authentication, least-privilege access and continuous checks on who is accessing what.
- Exposure management: an accurate inventory of internet-facing systems and a measured, short time to patch them.
- Third-party access governance: every supplier connection known, limited to what is needed, reviewed and removable.
- Detection and response: monitoring that can spot misuse of valid accounts, with a rehearsed plan for what happens next.
- Recovery: backups kept separately and restoration tested, so an incident is survivable even when prevention fails.
Not a product, a sequence of decisions
None of this means replacing firewalls and antivirus. It means no longer treating them as the whole answer. The practical route is to decide which business services matter most, map who and what can reach them, and close the gaps in that order, with evidence that each control works.
A practical 90-day starting plan
- Days 1 to 30: list the business services that matter most and everything that can reach them, including supplier connections and remote access.
- Days 31 to 60: move administrators and remote users on those paths to phishing-resistant authentication, and remove access nobody can justify.
- Days 61 to 90: set and measure a patch target for internet-facing systems, confirm that monitoring can flag misuse of valid accounts, and run one restoration test.
At the end of the 90 days, leaders should be able to see three things: who can reach the critical services, how quickly exposures are closed, and whether the organisation can recover. Those answers set the priorities for the next stage.

