Ask ChatGPT whether secure web hosting exists and it will tell you, with total confidence, that secure web hosting is "absolutely a thing." Then it lists features: SSL, a firewall, DDoS protection, malware scanning, backups.

I've been building websites since 1999 and running a WordPress hosting company since 2011, and I disagree. There's no such thing as a secure host. There are hosts that do the work, and hosts that sell the label.

The short answer: "Secure WordPress hosting" describes a set of features, not an outcome. No host can make a WordPress site secure, because the most common ways in (stolen logins, hijacked sessions, vulnerable plugins) sit where a host's firewall can't see them. What separates a good host is how fast it patches, how well it isolates your site, whether it keeps backups somewhere a hacker can't reach, and whether a person acts quickly when something gets through. Ask for proof of all four.

I'm Shane Larrabee. FatLab hosts and manages more than 200 WordPress sites, mostly for nonprofits, advocacy groups and small businesses. This post is the start of a series on website security, and I want to begin with the claim the whole hosting industry leans on, because the gap between what "secure hosting" promises and what it delivers is where most of the damage happens.

What "secure hosting" means in the hosts' own fine print

Every big host markets security. One managed WordPress host's security page calls its protection "bulletproof." Another tells you you're safe "100% of the time." Then read the terms, and the big hosts hand most of the job back to you.

WP Engine's terms of service say the customer must configure its use of the service to keep it secure, including installing patches, and that WP Engine is "not responsible for any loss or harm suffered by you" in a security incident. They also say WP Engine doesn't guarantee against data loss and that you should keep your own offline copy of your site. Kinsta's security roles page is more direct: the host owns the network and infrastructure, and "you are responsible for installing, updating, and removing any plugins and themes."

I'm not criticizing either company for this. It's honest, and it's how the industry works. The security world calls it shared responsibility: the host secures the server, you secure what you put on it. WordPress's own hardening guide says the same thing, that a host provides security only up to a point, and opens with a line I'd put on every hosting sales page: "security is not about perfectly secure systems." The problem is that the marketing page says "secure WordPress hosting" and the contract says "mostly your job," and most site owners only ever read the marketing page.

A website owner reading a hosting company's terms of service at a desk next to a laptop

"A host can promise you a secure server. Nobody can honestly promise you a secure website, because the website is the part you, your staff and your plugins touch every day."

What a host's firewall can and can't stop

Firewalls are the feature hosts lean on hardest, so it's worth knowing what they actually catch.

In 2025, Patchstack tested the defenses of five hosting providers against 11 real WordPress plugin vulnerabilities, using plain public exploit code with no tricks. The hosts' firewalls and scanners, as sold to ordinary customers, stopped about 12% of the attacks. The best of them, a host using Cloudflare's firewall, blocked 4 of 11. Two blocked none. Patchstack sells a competing product, so weigh that accordingly, but the pattern makes sense: a generic firewall is good at generic attacks and weak against an exploit written for one specific plugin, where the malicious request looks like normal traffic.

Our own numbers show both sides of that. Every production site we host sits behind Cloudflare's enterprise firewall. In the three days before I wrote this, it blocked 334,680 requests across 147 of our sites, about 111,000 a day, and 272,010 of them matched known exploit signatures. Every one of the 147 sites had at least one attack blocked. That's a lot of attacks that never reached a server. It's also not a guarantee, and I wouldn't let anyone tell you it is. (Those figures come from Cloudflare's own security log for three days, the most it keeps, and may be sampled.)

Hosts know this, which is why the better ones are now adding protection aimed at specific plugin flaws on top of the general firewall. Kinsta added Patchstack's protection to every plan this month, and to its credit, its own product page says that doesn't cover every possible vulnerability and that updating is still essential. That's progress. It's still a layer, not a promise. For how the layers fit together, see security plugins vs. server-level protection.

Three breaches no host feature would have stopped

The best evidence I have isn't a study. It's our own ticket history. Here are three real cases from the past six months, anonymized.

1. The first breach of a site we host

In August 2026, a medical mission nonprofit we host emailed us that their homepage had been replaced with a gambling page. It hadn't been, exactly. Real visitors saw the normal site. Google's crawler was being served a casino affiliate page, and the cloaking code was built to show the clean site to almost every security scanner and monitoring tool. The only reason anyone saw it at all is that our CDN cached the crawler's version for a while.

The way in wasn't a server flaw or a plugin exploit. Five days earlier, a new administrator account had been created on the site from inside a staff member's own logged-in session, on her own connection, in the middle of her normal editing. Whatever rode along on that session, it had every permission she had. No firewall can tell the difference between you and something using your session, because as far as the server knows, it is you.

The infection was out 21 minutes after the ticket came in, and Google re-crawled a clean homepage about 90 minutes after that. We weren't perfect: a second pass the next day, built from the server's access logs, found a file manager plugin the attacker had installed that the first sweep missed. That's the lesson we kept: clean from the logs, not from what you can see on disk. The client wasn't charged, two-factor authentication became a free option for every client, and we stopped saying "zero breaches." What's still true is that none of our servers has been breached. That's the honest claim, and it's the only one any host should make.

A staff member editing a WordPress website on a laptop while logged in, the kind of session no hosting firewall can tell apart from an attacker

2. A "secure" platform that got breached anyway

A national nonprofit came to us after being hacked on a well-known enterprise WordPress platform, one whose live environment is read-only, which is a genuinely strong protection. Someone had been guessing the password of the built-in admin account since early May, slowly enough that the lockout never fired (brute force attacks rarely look like the movies). They got in at the end of July.

The read-only live site did its job: over a hundred attempts to install code there were rejected. So the attacker wrote to the development copy instead, and a routine maintenance deploy carried ten fake plugins, nine of them backdoors, onto the live site. The platform's design was secure. The password wasn't, and the deploy process trusted whatever was in development. We took a copy off the platform, cleaned it on an isolated server, checked it four independent ways, and had them live on our hosting three days after the first call.

3. One flaw, three hosts, three breaches

In July, WordPress fixed a pair of flaws in its core software that together let an attacker take over a site without logging in. The fix shipped on July 17. By July 19, two sites we later worked on had rogue administrator accounts: one on an enterprise platform, one on a large mainstream host. A third, on a server in the owner's home, was still on an unpatched version in late August, when bots created 40 administrator accounts on it in a month. None of them were on our hosting.

On two of the three, the fix existed and simply hadn't been deployed: on one it sat in a development copy that never went live, on the other the way the site was built blocked dashboard updates. On the third, the host cleaned out the malicious files but left the attacker's admin accounts in place, so the attacker could have walked straight back in. Three different hosts, the same flaw. The outcome depended on how fast someone applied the update and how thorough the cleanup was, not on which host had the better security page.

It isn't only our cases. In October, attackers stole the connection keys of UpdraftCentral, a popular service for managing many WordPress sites from one dashboard, and used them to create admin accounts and plant hidden plugins on connected sites. No plugin was vulnerable, and no host firewall could have told those logins from the real thing.

"Every breach I've cleaned up came in through a door the host's security page doesn't mention: a password, a session, a plugin nobody updated."

What actually separates a good host

If "secure" is the wrong question, here's what I'd measure instead. None of these make a site unhackable. All of them make a breach less likely, smaller, and shorter.

What matters What to look for What we do
Patch speed Server software (PHP, the web server, the operating system) patched within hours to days of a critical fix, with dates to prove it PHP's critical October 2026 fix was running on our servers about 10 to 17 hours after release; November's nginx fixes within about a day
Plugin and core updates Someone actually updates WordPress and its plugins, on a schedule, and checks that nothing broke Weekly updates, tested on a copy of the site first
Isolation Your site can't be reached through someone else's Our servers are shared only with other FatLab clients, never resold shared hosting
Backups out of reach Copies kept off the server, so a breach can't take them too Daily backups kept for 30 days, on the server and off it
A person who acts Someone notices, answers and fixes, fast, and the cleanup isn't an upsell Cleanup is part of the service; the breach above was out in 21 minutes

A stopwatch beside a server rack, representing how fast a WordPress host patches critical vulnerabilities

Patch speed is the one I'd press hardest on, because it's the one that's changed the most. More than 100,000 vulnerabilities have been published in 2026, more than double last year. For WordPress specifically, Patchstack counted 11,334 new vulnerabilities in 2025, up 42%, and found that about half of the high-impact ones were being exploited within 24 hours. When the window is that short, "we patch monthly" means a month of exposure. Verizon's 2026 Data Breach Investigations Report found that exploited vulnerabilities passed stolen credentials as the most common way into a breach, at 31%, for the first time in the report's 19 years.

Buyers are starting to notice. When a nonprofit came to us in November, the reason was simple: their previous host had taken a week to patch a critical PHP flaw. A week is a long time when exploit code shows up in days. For the full story on server patching, see what the Apache and nginx vulnerabilities mean for your WordPress site.

My forecast: more holes, found faster

Here's my opinion, and I'll label it as one. I expect the number of discovered vulnerabilities, and the number of attacks using them, to keep climbing for the next few years. AI tools are now finding real, serious bugs in mature software. Google's threat intelligence team counted 141 exploited vulnerabilities in the first eight months of 2026, more than in all of 2025.

To be fair to the other side: most of those 100,000 bugs will never be used in an attack. VulnCheck found the share that gets exploited has actually fallen, to about 1.4% in the first half of 2026, and some researchers think the AI angle is overhyped. I take that seriously. A lot of the growth is more bugs being counted, not more danger. But the number of vulnerabilities actually used in attacks still went up, and the ones that matter are being used faster.

"AI has also enabled the good guys to act so much faster."

That cuts both ways, and it's why I'm optimistic about hosts that do the work. The same tools that find the holes help us find out which of our servers are affected, check versions across the fleet in minutes, and answer every worried client the same day. The hosts that will struggle are the ones still treating security as a feature list. I hope I'm wrong about the volume. I don't think I am.

Five questions to ask any host

Don't ask a host "is your hosting secure?" Every one of them will say yes. Ask these instead, and listen for dates and specifics.

A small business owner on a video call with a web hosting provider, asking about security before choosing a host

  1. "When was the last critical server vulnerability, and how many hours did it take you to patch it?" A good host has a number. "We're always on top of it" isn't one.
  2. "Who updates my plugins and WordPress core, how often, and what happens if an update breaks something?" If the answer is "you do," you're buying a server, not security.
  3. "Where are my backups stored, and could a hacker who gets into my site delete them?" Backups on the same server as the site aren't enough. More on that in backup plugins vs. server-level backups.
  4. "If my site is hacked tonight, what happens in the first hour, who does it, and what does it cost?" You want a person, a process, and no surprise invoice.
  5. "Who else is on my server, and what keeps their problem from becoming mine?" Shared hosting is fine when you know who you're sharing with.

Our answers are in the table above. If you're comparing hosts more broadly, our guide to choosing WordPress hosting covers the rest, and if a client's IT department is asking you these questions, here's how to answer a vendor security questionnaire.

What you can do that no host can

A host can only protect the server. These are the doors only you can lock, and they're where most of the breaches above actually started:

  • Turn on two-factor authentication for everyone who can log in. It's the single best fix for stolen passwords and slow password guessing. Here's what 2FA stops and what it doesn't.
  • Get rid of the admin username and any account nobody uses anymore.
  • Remove plugins you don't need, and only install ones that are actively maintained. Here's how WordPress plugins get compromised.
  • Know who to call. If you don't know whether your host or your developer handles a hacked site, find out before you need to.

If you'd rather not manage any of this yourself, that's what our managed WordPress security does: the server, the updates, the monitoring and the cleanup, handled by one team. And for a wider view of the threats themselves, see how WordPress sites actually get hacked.

Frequently Asked Questions

Is there such a thing as secure web hosting?

Not in the sense the phrase suggests. Hosting can be well secured at the server level, with fast patching, a firewall, isolation and off-server backups, but no host can make a website secure, because many attacks come through logins, sessions and plugins the host doesn't control. Look for a host that does the security work and can show you how, not one that promises you're safe.

What is the most secure WordPress hosting?

There's no single most secure host. The most secure setup is one where the server is patched within hours to days of a critical fix, WordPress and its plugins are updated weekly and tested, backups are stored off the server, every login uses two-factor authentication, and someone responds fast when something goes wrong. Managed WordPress hosting that includes updates and cleanup covers more of that than hosting that only runs the server.

Does managed WordPress hosting protect my site from being hacked?

It lowers the risk but doesn't remove it. Most managed hosts secure the server and the network. Many still leave plugin updates, user accounts and passwords to you, and their terms of service usually say so. Ask exactly which of those the host handles before you assume you're covered.

If my website is hacked, is my web host responsible?

Usually not, unless the host's own server or infrastructure was breached. Most hosting agreements make the customer responsible for plugins, themes, passwords and user accounts, which is where most hacks start. Check your host's terms of service, and find out whether cleanup is included in your plan or billed separately.

Is a web application firewall enough to protect a WordPress site?

No. A firewall blocks a large volume of known and automated attacks, but in Patchstack's 2025 test, hosting firewalls stopped only about 12% of exploits against real WordPress plugin vulnerabilities. A firewall buys time. Updates close the hole, and two-factor authentication protects the logins a firewall can't judge.

Is WordPress safe from hackers?

WordPress core is well maintained and usually not the problem: Patchstack found only six vulnerabilities in core in 2025, all low priority, against more than 11,000 in plugins and themes. Core isn't immune, though: CISA added four WordPress core flaws to its Known Exploited Vulnerabilities catalog in 2026, and the sites that got hit were the ones that hadn't updated. A WordPress site is as safe as its plugins, its passwords and how quickly it gets updated.