Is AI Killing WordPress?

Why the question feels urgent

Start with what a small business can do today. Take out a subscription to ChatGPT, Copilot, Perplexity, Cursor, or a combination of them. Describe the business, ask for a site structure, generate the copy and images, produce the code, and get a convincing first version online in minutes.

From the buyer's side, that can make the old website process look difficult to justify. Why wait weeks for a proposal, design, and build when a prompt can produce something immediately? Why pay an agency for a standard brochure site? Why install a CMS at all if AI can generate every page?

The scale of WordPress makes those questions more significant. As of 27 July 2026, W3Techs reports that WordPress runs 41.2% of all websites and has 59.1% of the market among sites whose content management system is known. What happens when a technology capable of generating websites in minutes meets a platform used by such a large part of the web?

Does AI replace the platform, reduce the need for professional WordPress work, or simply move the difficult part of the job somewhere else? And does an impressive first version tell us anything about how the website will be managed a year later?

What AI may actually kill is the commodity build

There is a real disruption hiding inside the exaggerated headline. A five-page brochure site built from a familiar template has become easier to produce. AI can draft the copy, suggest a structure, generate images, write CSS, and assemble standard components. A client who only sees the finished pages may reasonably ask why the old process still costs what it did.

That pressure will be felt by freelancers, junior developers, and agencies selling implementation hours without a deeper service. AI does not have to replace every developer to change hiring. It only has to let one capable person complete more routine work, or let a small business defer professional help until the site becomes complicated.

The response cannot be to pretend the tools are useless. It is to stop presenting page assembly as the entire value of a WordPress professional. Discovery, information architecture, conversion strategy, accessibility, performance, custom integrations, security, migrations, quality assurance, and long-term support were always the harder work. AI simply makes the difference more visible.

A business website still needs an operating system after launch

The first version of a website is a small part of its lifetime cost. The business then hires someone new, publishes a campaign, changes a service, replaces a team member, creates a landing page, fixes an old redirect, responds to a security issue, and updates a privacy notice. Each task needs a safe and repeatable path.

This is where the phrase "AI website" becomes vague. Does it mean AI generated the original code? Does an AI assistant remain responsible for every change? Is there an editor behind it? Where are the content, images, redirects, user permissions, form submissions, and revision history stored? What happens when the model, prompt format, or generation service changes?

A content manager should not need to understand a component tree to publish a case study. An SEO specialist should not need a developer to change a canonical URL. A new employee should not have to learn the prompts that happened to generate the current navigation. These roles need a stable interface and defined permissions, which is precisely what a CMS provides.

AI editing needs a safety net

Asking AI to manage an existing page introduces a different risk from asking it to generate a new one. The model can hallucinate a fact, but it can also misunderstand the boundaries of the task. A request to shorten one paragraph might change the heading above it, remove a target keyword, rewrite a link, alter an image description, or update a reusable block that appears on several other pages.

These changes are particularly easy to miss because the result can still read naturally. The page may look polished while a product claim has become inaccurate, an internal link has disappeared, a legal qualification has been softened, or carefully approved brand language has been replaced. An AI system does not share the content manager's understanding of which words are negotiable and which exist for commercial, legal, accessibility, or SEO reasons.

Prompts are instructions, not reliable change boundaries. Telling an assistant to "only edit this sentence" is useful context, but it is not the same as technically restricting write access to that field. A dependable workflow needs controls around the model.

WordPress already has useful foundations for this approach: revisions, drafts, previews, user roles, and granular capabilities. AI can sit inside that governed workflow as an assistant. Replacing the workflow with unrestricted prompt-based editing would remove the very safeguards a content team needs most.

The hidden cost of inventing a CMS for every company

A generated custom website can look cheaper because its first invoice excludes the organisational cost that appears later. Once non-technical staff need to manage it, the company has three choices.

1. Keep a developer or AI operator in the loop for routine changes. This turns ordinary publishing into a support request and creates a queue around work a content editor could have completed directly.
2. Build an administration layer. The company must define content models, authentication, roles, previews, revisions, media handling, redirects, backups, migrations, and security. It has started building a private CMS, whether it uses that name or not.
3. Adopt an existing CMS. The generated front end can remain custom, but a mature content platform becomes the source of truth behind it.

The second option can be correct for a product company with unusual requirements and an engineering team. It rarely makes sense for an ordinary business website. Bespoke systems also create bespoke training. Every agency handover and every new employee must learn an interface that exists nowhere else.

WordPress spreads that cost across an ecosystem. Editors bring transferable knowledge. Agencies can take over from one another. Hosting, backup, security, analytics, forms, commerce, multilingual publishing, and SEO all have established solutions. The open-source licence also means the business is not renting access to its publishing system from a single AI website vendor.

AI is entering WordPress itself

The premise that businesses must choose between AI and WordPress is already out of date. WordPress formed a dedicated AI team in 2025, and WordPress 7.0 moved the first important foundations into the platform. The WordPress 7.0 release introduced AI infrastructure intended for core, plugin, and workflow development.

This direction matters more than adding a chat box to the editor. It treats AI as infrastructure. A plugin developer can build a feature against WordPress capabilities and a common AI client while the site owner retains a choice of provider. That is a healthier long-term model than making the whole website dependent on whichever generation service is fashionable this year.

WordPress is stable, but stable does not mean static

Stability is sometimes mistaken for a lack of innovation. For a business platform, stability is the fact that content written years ago can still be edited, common skills remain useful, and a company can change its host or agency without rebuilding its entire publishing operation.

The WordPress enterprise programme describes the platform being used at scale across media, e-commerce, higher education, and content marketing. Large organisations do not choose a CMS only because it can render a page. They choose around governance, extensibility, security processes, integration options, recruitment, support, and total cost over time.

WordPress can modernise without asking those organisations to abandon the content and workflows they already own. The block editor changed page creation. REST APIs made headless implementations practical. The AI Client, Connectors API, and Abilities API now create another integration layer. The durable part is not a particular editor or model. It is the content system and the compatibility contract underneath them.

What changes for WordPress developers and agencies

WordPress work will not disappear, but its centre of value is moving. Clients are less likely to pay a premium merely because a professional can turn a design into standard pages. They will still pay to reduce risk, connect systems, improve outcomes, and keep a revenue-producing website healthy.

AI can also make genuinely custom WordPress work more practical. A developer can use it to scaffold a theme, create block templates, draft theme.json settings, build component variations, refactor repetitive code, and generate the first version of tests and documentation. Work that once consumed the opening hours of a project can become a starting point for review and refinement.

That speed can create more room for creativity rather than less. A developer can explore several layout directions, prototype an interaction, test a different content pattern, or discard a weak idea before the client has committed to it. The result does not have to be another generic AI-generated site. Used well, AI can make a more tailored theme affordable within the same project constraints.

Faster output is only a benefit when it survives professional review. The developer still has to understand the generated code, remove unnecessary complexity, check accessibility and performance, test different devices and content states, and make sure the client receives a maintainable system. AI can compress production time. It cannot take ownership of the result.

• Design an intentional AI workflow. Decide where AI helps with specifications, custom-theme scaffolding, code, tests, documentation, and repetitive content work, and where human review is mandatory.
• Sell decisions and outcomes. Content modelling, accessibility, performance budgets, analytics, conversion paths, security boundaries, and integration architecture are more defensible than a bill based on how long a page took to assemble.
• Own the years after launch. Maintenance, monitoring, backups, vulnerability response, editorial support, training, SEO changes, and continuous improvement create more durable value than the initial build.
• Build model-neutral features. WordPress's new AI foundations let plugin developers support capabilities without permanently coupling their product to a single provider.
• Make governance part of the feature. AI workflows need capability checks, human approval for consequential actions, cost controls, and a record of what changed.

The job becomes less about being the person who knows where a setting lives and more about being the person who can design a reliable system. That is a harder role to automate and a more useful role for a client.

An AI-generated site can be right, and WordPress can be right

Not every project needs WordPress. The sensible choice depends on the operating life of the site, not on loyalty to a tool.

There is also a hybrid answer. AI can help create blocks, layouts, content, code, and operational automations while WordPress remains the source of truth. Businesses get faster production without asking their staff to abandon a governed publishing environment.

The verdict: AI will change WordPress because WordPress is useful

AI is not killing WordPress. It is challenging the price of routine implementation and exposing how little of a website's value lives in the first generated version. That will be uncomfortable for anyone whose offer ends at "we build WordPress websites".

WordPress remains valuable for the less glamorous reason that businesses need continuity. Content managers need to publish. SEO specialists need to control the details. Agencies need to maintain many sites. Companies need permissions, integrations, portability, and people they can hire. Recreating those capabilities for every AI-generated website would be expensive, repetitive, and difficult to govern.

The likely future is not AI or WordPress. It is AI working through WordPress, with developers and agencies responsible for turning fast generation into a dependable business system.

Common questions about AI and WordPress

Sources

• W3Techs: WordPress usage and CMS market share
• WordPress 7.0 release announcement
• WordPress Core: Introducing the AI Client
• WordPress Core: Introducing the Connectors API
• WordPress Developer Blog: Introducing the WordPress MCP Adapter
• WordPress for enterprise