Opens in a new tab

Website Security Updates: Install Plugin Fixes in Days, Not Weeks

How fast to install website security updates: an exposure check for plugins and servers, a deadline in days, and what to do when no fix exists yet.

A cracked stone sphere held together by a clear glass patch, as figures race toward it from both sides

Security fixes for your website, its plugins and anything else people outside the company can reach should go on within days of release, or install automatically with a restore you have already tested. Tools only your staff can reach can wait for your regular update cycle, and a plugin with no fix yet needs a different answer: remove it, disable the affected feature or put protection in front of it.

A cracked stone sphere held together by a clear glass patch, as figures race toward it from both sides
In this article · 6 sections
  1. 01Why the window between fix and attack is short
  2. 02An exposure check you can run this week
  3. 03Set the deadline by what each row exposes
  4. 04When a plugin has no fix yet
  5. 05Update freezes during busy periods
  6. 06What a fast update rule does not cover

The deadline matters because a published fix shows attackers where the hole was. On 17 November 2026, Google Threat Intelligence Group reported that for the first time it had identified attackers using large language model frameworks to find and exploit flaws with no person in the loop, by comparing popular web software before and after its fixes.

Why the window between fix and attack is short

Google’s report says attackers may find it more efficient to work on patched flaws, because they can focus on “a finite set of source materials”, such as the code before and after the fix and its release notes. The framework it found compares code before and after patch announcements, drafts an exploit and runs it without human intervention. Victim data points to French websites in media, entertainment and education, and the report gives no figure for how often such tools reach smaller companies. The report, checked on 19 November 2026, says these tools generate forensic traffic that makes detection more likely, yet they “remain highly capable of achieving operational objectives”, so the noise protects only a team that reads its server logs.

Before any of this, Mandiant found that 29% of the n-days it tracked in 2023, flaws first attacked after a fix was available, were exploited within a week. Patchstack, which sells virtual patching, puts the weighted median time to first exploit at 5 hours for the most heavily exploited WordPress vulnerabilities of 2025, and its sensors picked up exploitation attempts against a WordPress core flaw roughly 90 minutes after WordPress 7.0.2 shipped the fix. Meanwhile the median time to patch in Verizon’s 2026 Data Breach Investigations Report, as reported by Help Net Security, rose to 43 days from 32, on data that predates the surge in AI-assisted vulnerability research.

An exposure check you can run this week

Start with a list of everything someone on the internet can reach: the website’s core software, each plugin and theme, any booking or form tool you host yourself, remote-access or file-sharing servers, and any other web application on your own server. Tools a vendor hosts for you, such as an email marketing service, go on the list as well, marked as the vendor’s to patch.

For each row, fill in five things: who patches it, the one person who owns its updates (at an agency, a named contact who receives the update alerts), whether automatic updates are on, how many days a published security fix for it has been waiting, and whether anyone has actually restored a copy of it. A backup job that reports success is not a tested restore. The waiting days come from the update notice or the release date in the plugin’s changelog; if nothing is waiting, write 0.

Here is the check filled in once, for a hypothetical company with a WordPress site and one file-sharing server. The numbers are illustrative, not from a client.

SystemWho patches itUpdate ownerAuto-updateDays a published security fix has waitedRestore tested
WordPress coreUsAgencyOn0Yes
Booking pluginUsNobody namedOff19Yes
Contact form pluginUsMarketing managerOff6Yes
ThemeUsAgencyOff0Yes
File-sharing serverUsNobody namedOff41No
Email marketing toolVendorVendor, update practice confirmedNot oursNot oursNot ours
Total, systems we patch53 with a named owner1 on664 of 5

In this example the two rows with nobody named hold 60 of the 66 days that published fixes have waited, and the file-sharing server alone holds 41. Name the owners before you argue about deadlines. Copy the table and replace the rows with your own.

Set the deadline by what each row exposes

Once the list exists, each row falls into one of four cases.

RowRule
Internet can reach it, you patch it, a fix existsAutomatic updates with a tested restore, or the named owner installs the fix within your written number of days
Internet can reach it, you patch it, no fix yetRemove it, disable the affected feature, or put protection in front of it until a fix ships
A vendor hosts itConfirm the vendor updates it, and record the answer
Only staff can reach itBatch it into your regular update cycle

For the first case, the answer to how often to update plugins and other internet-facing software is a deadline: a number of days you pick and write down. The timelines above put it in days, not weeks, and the clock starts when the fix is published, which may be well before anyone at your company notices it.

Automatic updates can meet a deadline in days, with conditions. WordPress’s documentation says administrators opt in to auto-updates theme by theme and plugin by plugin, that WordPress runs them twice a day by default, and that depending on your server, installation or plugins the schedule may not work correctly. It also suggests making sure you can roll back to a previous version of the site before you switch them on. On an auto-updated site, the owner’s job is to confirm the updates ran and that the restore works.

CISA’s guidance for businesses points the same way. It advises enabling automatic updates where possible, prioritising critical vulnerabilities in public-facing or legacy systems, and setting up regular patching procedures and tests with your IT team. For vendor-hosted tools it advises confirming that your vendors regularly update their systems. Ask once, write down the answer, and put the vendor in the owner column.

An owner is one named person. An agency can own a row if the contract says so and a named person there receives the update alerts. Rows only staff can reach still get updated, on the regular cycle, behind everything the internet can touch.

When a plugin has no fix yet

Updates only help where a fix exists. Patchstack reports that 46% of the WordPress vulnerabilities disclosed in 2025 did not receive a fix from the developer in time for public disclosure. Its conclusion, that site owners can’t rely on plugin updates as a security measure, comes from a company that sells the alternative. The 46% still means your rule needs a branch for rows with nothing to install.

Work through that branch in order. If the plugin is not essential, remove it, and there is nothing left to patch. If the site needs the plugin but can run without the affected feature, disable that feature until the fix ships. If you need the feature as well, put protection in front of it, such as a web application firewall rule or a virtual patch from your host or a security vendor, and set a date to check for the fix again.

Update freezes during busy periods

If your team freezes website changes during a busy season, the freeze protects the checkout from a broken update and does nothing about an attacker who has read the fix. Write one line into the freeze: security fixes for internet-facing systems are exempt, and the owner takes a restore point before installing them. We know of no reliable figure for how often updates break sites; the restore point is what makes a bad update recoverable in the weeks you can least afford downtime.

What a fast update rule does not cover

A deadline does nothing for flaws attacked before any fix exists. In Mandiant’s 2023 data, which counted exploited flaws in software generally and not plugins in particular, the ratio of n-days to zero-days (flaws attacked before any fix existed) was 30:70. That is why removing plugins you do not need, and the no-fix steps above, matter as much as the deadline.

Share this article

Omar Ibrahim

Founder of Dr-Business. I help businesses turn AI into reliable operating systems - the workflows, guardrails, and judgment that make it deliver real results, instead of chasing prompts and hype.

More about Omar
Leave a comment

Your email address will not be published. Required fields are marked *

Newsletter

Sharper signal. Smarter decisions.

Join our newsletter for our best thinking on AI and systems, delivered straight to your inbox - no noise.

Subscription Form

No spam. Unsubscribe anytime.