“Zero-day” is one of the most overused terms in security reporting. It sounds dramatic, and it gets attached to almost any serious bug. Using it precisely matters, though, because the life of a vulnerability has distinct phases, and the defensive options in each phase are different.

The timeline of a vulnerability

Picture one flaw moving through time.

  1. Introduction. A developer writes the faulty code. It may sit unnoticed for years. Several famous flaws had existed for a decade or more before anyone spotted them.
  2. Discovery. Someone finds it: an internal tester, an independent researcher, a criminal group, or a government agency.
  3. Vendor awareness. The vendor learns about it, ideally through a private report.
  4. Patch release and public disclosure. Usually together: an advisory and a fix.
  5. Patch deployment. Customers actually install the fix, which happens gradually over weeks, months or never.

A zero-day is a vulnerability that is being exploited, or is known to attackers, before the vendor has released a fix. The name refers to defenders having had zero days to prepare. The window between step 2 (if the discoverer is an attacker) and step 4 is the zero-day period.

An n-day (sometimes called a one-day) is a vulnerability for which a fix exists but which remains unpatched on many systems. Here, “n” is the number of days since disclosure. Most real-world compromises involve n-days, not zero-days, because n-days are cheaper for attackers and there are far more of them.

Why the patch itself can raise the danger

It sounds backwards, but the release of a patch often increases the short-term risk for anyone who has not applied it. The patch is a precise map of what was broken. Skilled attackers can compare the old and new versions of a program, a technique known as patch diffing, and work out exactly which function changed and why. For many bugs, this leads to a working exploit within days, sometimes hours, of the fix appearing.

Before the patch, only the discoverer and the vendor knew. After the patch, everyone with the skill to read the change knows. Meanwhile, most organisations are still testing or waiting for their maintenance window. This is why “actively exploited” lists and emergency patch lanes exist: the gap between patch release and attacker use has shrunk to the point where monthly cycles can be too slow for internet-facing systems.

Coordinated disclosure: how researchers and vendors cooperate

When a well-intentioned researcher finds a bug, they face a choice. Publishing immediately alerts defenders but also arms attackers before a fix exists. Staying silent indefinitely leaves users exposed if the vendor never acts. Coordinated vulnerability disclosure is the compromise that has become the norm: the researcher reports privately, the vendor fixes within an agreed period, and details are published once a fix is available or the deadline passes.

Google’s Project Zero popularised a 90-day deadline, after which details are published whether or not a patch exists, with some allowance when a fix is close. Other researchers and organisations use different windows. Deadlines exist because, without them, some vendors historically sat on reports for years.

There are international standards for this process: ISO/IEC 29147 covers how vendors should receive and publish vulnerability information, and ISO/IEC 30111 covers how they should handle reports internally. A simple, widely adopted practice is publishing a security.txt file on a website (defined in RFC 9116), which tells researchers where to send reports. Many Indian companies still have no published way to report a vulnerability, so researchers end up emailing customer support, where reports get lost.

Bug bounties: paying for discovery

A bug bounty programme offers rewards for valid vulnerability reports within a defined scope. It formalises the relationship with external researchers: what is in scope, what testing is allowed, how to report, and what the payouts are. Large technology companies run them publicly; many smaller companies run private programmes through platforms that invite selected researchers.

Bounties work well for organisations that already fix known issues promptly. They work badly for those that do not. If a company launches a bounty while hundreds of basic issues remain open, it will pay repeatedly for findings it could have found with a scanner, and its developers will drown in duplicate reports. A vulnerability disclosure policy without payment, simply a safe and clear channel for reports and a promise not to take legal action against good-faith research, is often the better first step.

What defenders can actually do in each phase

Against true zero-days, patching is by definition impossible. Defence relies on things that do not depend on knowing the specific flaw: reducing what is exposed to the internet, network segmentation so one compromised system does not open everything, least privilege so a compromised application runs with limited rights, application allow-listing, and monitoring for unusual behaviour after compromise. Zero-days are expensive, and attackers tend to save them for valuable targets, so most organisations are more likely to meet them indirectly, for instance when a zero-day in a widely used file transfer or VPN product is used in a mass campaign.

In the days right after disclosure, speed matters most. Read the advisory, check exposure against your inventory, apply vendor workarounds if patching will take time, and watch threat intelligence for signs of exploitation.

For long-tail n-days, the issue is hygiene. Systems still vulnerable months later are usually forgotten assets, unsupported software or exceptions that never expired. Ageing reports, which list how long each vulnerability has been open, find these better than severity-sorted reports.

A note on language

Calling every serious bug a zero-day muddles priorities. If an incident report says “we were hit by a zero-day” but the vulnerability had a patch available for five months, the honest description is that an unpatched n-day was exploited. The first framing suggests nothing could have been done; the second points to a fixable process gap. Precise words lead to better lessons learned.

Written by the RP Strativa Team.

Leave a Reply

Trending