A zero-day vulnerability is a software flaw that attackers can exploit before the vendor has made a fix available, or before the vendor even knows about it. “Zero-day” refers to the number of days the defenders have had to prepare: none. A zero-day exploit is the code or technique that takes advantage of such a flaw, and a zero-day attack is its use against a real target.
People use the term loosely, and the looseness causes confusion. Many flaws that make headlines are not zero-days at all. They are known, patched flaws that attackers exploit against organizations that have not yet applied the update. Telling the two apart changes what you should do.
This page defines the term precisely, walks through the lifecycle of a zero-day, explains how it differs from an n-day, and sets out a practical response for the situations where no patch exists. It sits within our coverage of vulnerability management and exposure management.
Zero-day, n-day and known vulnerability
| Term | Meaning | Is a fix available? |
|---|---|---|
| Zero-day | Exploited, or exploitable, before a fix exists or before the vendor knows | No |
| One-day / n-day | A flaw that has been disclosed and usually patched, and is now being exploited against unpatched systems | Yes, but not yet applied |
| Known vulnerability | Any publicly catalogued flaw, with a CVE identifier | Maybe |
Here is the practical point. Most organizations are breached through n-days, not zero-days. A patch exists, and the attacker only needs the window between the patch release and the moment you apply it. Zero-days get the attention, and slow patching causes most of the damage.
That said, zero-days matter because the usual defense, installing the update, is unavailable.
The lifecycle of a zero-day
- Introduction. A flaw is written into the software. It may sit unnoticed for years. Recent PostgreSQL fixes included a bug reported to be 22 years old. See PostgreSQL’s 52 security fixes.
- Discovery. Someone finds it: a vendor engineer, an independent researcher, a criminal group or a government agency.
- Exploit development. A working attack is built. This takes skill and time, though AI-assisted tools are lowering the effort. See autonomous penetration testing.
- Exploitation in the wild. The flaw is used against targets. If this happens before the vendor knows, it is a true zero-day. Often it first appears as an unexplained intrusion that investigators later trace to an unknown flaw.
- Disclosure. The vendor learns of the flaw, through a researcher’s report or because an attack is detected, and assigns a CVE identifier.
- Patch or mitigation. The vendor releases an update, or guidance to reduce risk, such as disabling a feature.
- Listing and awareness. If the flaw is being exploited, CISA may add it to the Known Exploited Vulnerabilities catalog. Public proof-of-concept code often appears, and exploitation by many more attackers follows.
- Patch adoption. Organizations apply the fix over days, weeks or never. Exploitation continues against the unpatched.
The period between steps 4 and 6 is the zero-day window. The period between 6 and 8 is the n-day window, and for most organizations it is the longer and more dangerous of the two.
Google Project Zero tracks zero-days exploited in the wild and publishes its research, which gives a useful view of how many are seen each year and in which products.
Why zero-days are hard to defend against
- No signature. Detection tools that look for known malware or known exploit patterns may see nothing.
- No patch. The standard fix is unavailable until the vendor ships one.
- Often found in widely used, internet-facing products. Edge devices, gateways, remote access software and management tools are frequent targets because they sit at the boundary and are trusted.
- Short or no warning. You may learn about it from the news the day of the attack, or after the fact.
How to respond when no patch exists
1. Find out whether you run the affected product
The first question is the dullest and the most important: do we have it? This needs an up-to-date inventory of what you run, including software inside containers and dependencies inside applications. See software composition analysis and SBOMs.
2. Establish whether it is reachable
A vulnerable product on an isolated network is a different case from one on the internet. Identify which instances are exposed to the internet or to untrusted networks. See exposure management.
3. Apply the vendor’s mitigation
Vendors often publish interim steps: disable a feature, restrict access to a management interface, change a setting, or apply a configuration file. Follow the vendor’s advisory, and record what you changed so you can undo it after patching.
4. Reduce exposure yourself
If the vendor has not provided guidance, you can still shrink the attack surface:
- Remove the system from the internet, or put it behind a VPN or a zero trust access layer
- Restrict access by source address
- Disable or isolate the affected service
- Take it offline if its function is not essential
5. Limit what an attacker could reach if it is compromised
Assume exploitation is possible and ask what the compromised system can access. That means the credentials stored on it, the roles it can assume, the data it can reach and the systems it connects to. Reducing standing access and long-lived credentials limits the damage. See blast radius.
6. Increase monitoring
Watch the affected systems for unexpected processes, new accounts, outbound connections, and changes to configuration. Use the vendor’s indicators of compromise if provided. See incident response.
7. Hunt back in time
If the flaw is being exploited, attackers may have been present before the disclosure. Search logs back as far as you can for the indicators and for unusual activity.
8. Patch fast when the fix arrives, in order of exposure
When the update is released, apply it first to systems that are reachable and valuable. Use KEV and EPSS to order the rest.
9. Verify and document
Confirm the fix is in place, remove temporary mitigations if appropriate, and record the timeline for your audit and lessons-learned process.
Common misconceptions
- “We are not a target.” Zero-days are used by criminals as well as governments, and exploited at scale once public.
- “A firewall will stop it.” Many zero-days are in the products that sit at the boundary.
- “We will patch when the vendor releases it, so we are fine.” The window before a patch is exactly where you need other controls.
- “Zero-days cause most breaches.” Unpatched known flaws and stolen credentials cause far more.
How Cloudanix helps
Cloudanix does not discover zero-days or write the vulnerability lists. CISA maintains KEV, and vendors issue patches. What Cloudanix helps with is the work around them: knowing whether you are exposed, how reachable it is, and what it could lead to.
- Spotting your exposure when a flaw is added to KEV. Zero-Day Watch surfaces exposure across your workloads, container images and dependencies when a CVE lands on KEV or its EPSS score rises, and joins it to reachability and identity blast radius.
- Knowing what you run. Code Security generates a CycloneDX SBOM per repository and matches dependencies against a vulnerability database that includes CISA KEV and EPSS. Container image scanning covers build time and runtime for Kubernetes workloads. See which vulnerabilities are actually running in production.
- Reachability. CSPM shows which assets are internet-facing and ranks findings with contextual severity. The internet-exposed assets use case covers the reachability question.
- Reducing consequence. Cloud JIT and the other JIT surfaces remove standing privilege, so a compromised system reaches less. The attack path view shows what a compromised host connects to.
The boundary: Cloudanix cannot patch a vendor’s product for you, and it cannot see a system it does not know about. Static and dependency analysis plus runtime context is not a substitute for dynamic testing.
Further reading
- CISA Known Exploited Vulnerabilities Catalog
- Google Project Zero: 0-days in the wild
- FIRST: EPSS
- Ivanti Connect Secure zero-day on the KEV list: why exposure and blast radius matter
Frequently asked questions
What is a zero-day vulnerability?
A flaw in software that attackers can exploit before the vendor has released a fix, or before the vendor knows about it.
What is the difference between a zero-day and an n-day?
A zero-day has no patch available when it is exploited. An n-day is a disclosed flaw, usually with a patch, that attackers exploit against systems that have not applied it.
How can I protect against a zero-day if there is no patch?
Check whether you run the product, remove or restrict internet exposure, apply vendor mitigations, reduce what the system can reach, increase monitoring, and patch quickly when the fix is released.
Are zero-days the main cause of breaches?
No. Most breaches involve stolen credentials or known flaws that were not patched. Zero-days are important because the usual fix is unavailable.
How do I find out when a flaw is being exploited?
Watch the CISA KEV catalog and vendor advisories. EPSS gives a prediction of exploitation likelihood. Tools that match these against your inventory tell you whether it affects you.