Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
A fail-secure system protects the asset when something breaks; a fail-safe system protects people and keeps things moving. The choice decides what you lose on a bad day.

What Is the Difference Between Fail-Secure and Fail-Safe?
Systems break. The interesting question is what state they land in when they do. A fail-secure design defaults to denying access, protecting the asset at the cost of availability. A fail-safe design defaults to allowing access, protecting people and continuity at the cost of control. Neither is universally correct, and choosing without deciding is how organisations discover their answer during an outage.
Electrically operated locks make the distinction physical. A fail-secure electric strike needs power to release, so if the building loses power the door stays locked—good for a server room, dangerous on an escape route. A fail-safe magnetic lock needs power to hold, so a power cut unlocks it—good for a fire exit, useless for a cash office.
In the UK the choice is not purely a security decision. The Regulatory Reform (Fire Safety) Order 2005 places a duty on the responsible person to keep escape routes usable at all times, and BS 7273-4 sets out how door release mechanisms should behave when the fire alarm operates. Any access control design that holds a final exit door locked during an alarm will fail a fire risk assessment, whatever the security case for it. Practical installations resolve the tension with layers: fail-safe on escape routes, fail-secure on internal high-value rooms, plus green break-glass units, door monitoring, and alarms on every forced release.
Software engineers usually say fail closed and fail open, but the logic is identical. An authorisation check that cannot reach the policy service either denies the request (fail closed) or allows it (fail open). The OWASP principle of failing securely means the default path through an error handler ends in denial, and that exception handling never skips the check it was wrapped around.
The cost of getting it wrong in either direction is real. The CrowdStrike Falcon content update of 19 July 2024 crashed Windows hosts into a boot loop and grounded flights and payment systems across the UK; the endpoint refused to run rather than run unprotected, which is fail-secure behaviour with an enormous availability bill. Conversely, plenty of breaches have begun with a security appliance quietly passing traffic after its inspection engine died.
| Question | Points to fail-safe | Points to fail-secure |
|---|---|---|
| Can failure injure someone? | Yes | No |
| What does the system protect? | Availability of a service | Confidentiality or integrity of data |
| What is the regulatory driver? | Fire safety, health and safety | UK GDPR, PCI DSS, NIS Regulations |
| How long can the failure last? | Hours, with no manual alternative | Minutes, with a staffed fallback |
Map the decision against the CIA triad: fail-secure favours confidentiality and integrity, fail-safe favours availability. Then write the answer down. Most organisations have the behaviour buried in a vendor default rather than in a design document, which means nobody discovers it until the component fails.
In operational technology, fail-safe has a precise engineering meaning. A safety instrumented system defined under IEC 61511 trips the process to a designed safe state—usually de-energised and shut down—when it detects a fault. Security controls bolted onto that environment must not override the trip, which is why network security devices on an OT network are often deployed in monitoring mode, and why the Triton malware discovered in 2017 was so alarming: it targeted the safety controllers themselves.
Whichever default you choose, test it. Pull the power, block the policy server, expire the licence, and watch what happens. Our infrastructure and architecture review services include exactly that kind of failure-mode testing, so the behaviour on your worst day is one you chose.
Both, in different places. Final exit doors and escape routes should be fail-safe so that people can leave during a power cut or fire alarm, as UK fire safety duties require. Internal rooms holding servers, cash, or sensitive records are normally fail-secure, with door monitoring and break-glass release so that a locked door never traps anyone.
It depends on what the application handles. A payment or health service usually fails closed, because serving unfiltered traffic is worse than serving an error page. A high-volume content site often fails open to protect revenue. Whatever you choose, alert loudly on the condition, cap how long it may persist, and test the behaviour rather than trusting the default.
No. Fail-soft, or graceful degradation, means a system keeps running with reduced function when a component dies, such as a shop that accepts orders but defers fraud scoring. Fail-safe is specifically about the state the system lands in, prioritising safety and availability. A design can be fail-soft and fail-secure at the same time.
Test them. Disconnect the policy or identity provider, block the licence server, let a certificate expire in a staging environment, and record what each component does. Vendor documentation is often silent or wrong on this point. Capture the results in your architecture documents and revisit them after upgrades, because defaults change between major versions.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.