WordPress calls these updates improvements, which is a neat way of saying the platform changed under your feet. For anyone managing a live site, an automatic jump to a new version that quietly changes how content tools work feels more like someone moving the kitchen while the restaurant is still serving lunch. If the new AI connectors only function inside the block editor, every team that relies on Classic Editor, Elementor, Divi, Beaver Builder, or a custom workflow has just been handed a feature they cannot really use without changing how they work.
That small annoyance is the point. It exposes a bigger problem than WordPress itself. AI content is now being produced at a scale search engines and publishing platforms were never built to absorb. If content can be produced in bulk, tuned in bulk, and published in bulk, then the old assumptions around editorial effort, ranking signals, and platform compatibility start to look fragile fast.
WordPress broke the easy story
The practical issue with the WordPress 7.0 update is not that AI tools exist. It is that the tools arrive with a catch that turns them into a partial feature, not a usable one. If they are locked to the block editor, then anyone working outside Gutenberg is excluded from the benefit unless they change editor, remodel their workflow, or add another external tool into the mix.
That is not a minor implementation detail. It changes who can use the feature at all.
A lot of site operators in South Africa are not building from scratch on a clean, modern stack. They are running old but functional setups that have accreted over years. A marketing team may already have one process for drafting posts, another for sign-off, and a page builder sitting on top of a theme that only behaves if nobody pokes it too hard. The Classic Editor might still be there because the team knows exactly how it behaves. A builder like Elementor may be there because the agency that built the site handed over a working system and moved on.
Now the AI connector lands, but only if content is edited in a specific environment. That means the site owner gets to choose between three bad options.
One, abandon the new AI feature and carry on as before. Two, retrain staff and switch the editorial workflow over to blocks. Three, bolt on another AI tool outside WordPress, which defeats the promise of integrated convenience.
None of those options is elegant. All of them cost time.
The workflow tax is the real story
Every platform update gets sold as simplification until it collides with how people actually work. The moment a feature only functions inside one editor, you stop dealing with product messaging and start dealing with workflow tax.
That tax shows up in boring places first. Meta descriptions still need to be written. Alt text still needs to be checked. Outlines still need to be reviewed. Product pages still need to match the brand tone. If the AI connector was supposed to help with those jobs but only works in one interface, the team either moves the work into that interface or loses the benefit altogether.
For smaller South African businesses, this is where the economics get messy. Retraining content staff is not free. Reworking a theme so it plays nicely with blocks is not free. Paying a developer to untangle compatibility problems with a custom post type or an older plugin is definitely not free. Even a few hours of unnecessary downtime can sting when the site is doing lead generation, ecommerce, or support.
The people who designed the update probably see a clean product decision. The people who have to publish 20 posts a month on a mixed stack see friction. That gap between design and operation is where most real-world software failures live.
Block editor only is a hidden segmentation strategy
There is a second layer here that is easy to miss. When a platform reserves the useful new feature for one editing mode, it quietly creates a class system inside the CMS.
Users in the block editor get access to the latest AI tooling. Users elsewhere do not.
That matters because it splits the platform into modern and legacy camps, but not on any honest operational basis. It splits them on migration willingness. If you have the time and budget to adapt, you get the shiny new tool. If you do not, you get to watch the feature exist in screenshots.
For agencies managing several client sites, that split becomes operational debt. One client on blocks gets a different process from another client still on a builder. Content quality can drift. Tone can drift. SEO work can drift. A team member jumping between sites ends up maintaining three workflows for the same kind of task.
That is how inconsistency creeps in. Not through grand strategy failures, but through tiny compatibility decisions that make one site easy to update and another site awkward to touch.
The scale problem is not hypothetical anymore
The pitch behind the WordPress update asks a sharper question than most people realise. If the web doubles every year, did Google really understand what that would mean once AI content generation became cheap and fast?
That is not a polite academic question. It is the whole problem in plain language.
Google spent years building systems around human publishing patterns, even when those patterns were noisy. Editors wrote. Teams reviewed. Agencies charged by the hour. Content velocity had limits. AI has put a dent in that rhythm, because one operator can now spin up large volumes of text with far less effort than the old publishing stack assumed, while human review still decides what gets published.
Google’s own behaviour over the last few years shows the strain. Early in 2023, the company moved to clarify that AI-generated material is not automatically disqualified if it is genuinely useful to readers. By March 2023, the message had shifted toward people-first content rather than content judged only by how it was made. By March 2024, Google was talking much more aggressively about unhelpful content, spam, and large-scale content created mainly to rank.
That is not a company speaking from a settled theory. It is a company adjusting in public to an environment that moved faster than its original assumptions.
Search is now fighting volume, not just quality
The old SEO game was already messy, but at least scale had a cost. A content farm still needed people, time, and some level of coordination. AI has made bulk production cheap enough that mediocrity can flood the market.
That changes what search engines are up against.
If a query about insurance, solar power, payroll software, or small business marketing can return dozens of nearly identical pages written in slightly different tones, the ranking problem is no longer just about identifying quality. It is about filtering noise that arrives in industrial quantities. Google has said repeatedly that it cares about quality rather than the tool used to produce the content, but the real difficulty is obvious: detecting intent, originality, and usefulness at the scale AI now makes possible is a different class of problem.
The March 2024 core update was a sign of how serious the cleanup has become. Google tied it directly to unhelpful and spammy content, including pages built at scale to manipulate rankings. The company also talked about reducing unoriginal content materially. That alone tells you the volume of low-value output had become large enough to warrant a broad response, not just isolated manual action.
This is where the WordPress story stops being a CMS quirk and starts looking like a warning label.
Google is not just ranking pages, it is policing production volume
A search engine used to judge what was published. Now it also has to infer how the content factory operates.
That is a different job.
If a site suddenly starts producing hundreds of similar articles, the machine has to decide whether it is seeing a serious publishing operation or a synthetic content engine trying to game the index. If the pages are well written enough to avoid obvious spam signals, the judgement becomes harder. If the content is generated with prompt chains, edited lightly, and published at speed, the surface quality may be good enough to make detection less straightforward.
Danny Sullivan has repeatedly framed the issue in terms of quality rather than origin, which is sensible, but it also exposes the bind. Search systems can say they do not care whether a human or a model wrote the text. They still have to decide what deserves visibility when thousands of acceptable-looking pages are being pushed into the index every day.
That is the scale challenge Google may have underestimated. Not the existence of AI content, but the sheer amount of it, and the speed at which it can be made, revised, localised, and republished.
South African businesses will feel this in ordinary ways
This is not just a Silicon Valley problem dressed up as a thought piece. South African businesses are already operating in an environment where budgets are tight, SEO is under pressure, and site maintenance often happens in bursts rather than as a clean engineering process.
A company in Johannesburg, Cape Town, or Durban with a lean marketing team is far more likely to keep an older workflow running than to rebuild its content pipeline just to satisfy a platform update. That means feature mismatches will keep happening. New tools will land inside interfaces many teams do not want to use. AI features will be marketed as productivity gains while the real cost sits in retraining, migration, and cleanup.
Meanwhile, the content race gets uglier. If competitors can generate ten variations of the same article faster than your team can brief one decent one, the temptation is to match volume with volume. That is how feeds fill with sameness. That is how search results get clogged with pages that technically answer a query but never really help anyone.
For local publishers, agencies, and in-house teams, the better response is not to chase output counts. It is to build tighter editorial control, clearer site architecture, and cleaner workflows around the content that actually matters.
A practical response beats a panic response
There are a few sensible moves here, and none of them require pretending the problem is abstract.
1. Audit where your WordPress content actually gets created. If the block editor is not part of the real workflow, do not assume a block-only AI feature helps you. 2. Check which parts of the site rely on legacy builders, custom fields, or older plugins. The more patchwork the stack, the more likely a forced editor shift will create problems. 3. Decide which AI tasks are genuinely useful. Meta descriptions, outlines, internal link suggestions, image alt text, and summarisation are practical. Bulk article generation for its own sake is usually a mess. 4. Watch content quality per page, not just page count. A site with 40 strong pages will beat a site with 400 weak ones more often than people like to admit. 5. Build review steps that catch tone drift and duplication before publication. AI makes consistency easier to lose if no one is steering it.
If the WordPress update proves anything, it is that integrated AI is only useful when it fits the way people already work. If it forces a complete workflow change, it stops being a tool and starts being a migration project.
The uncomfortable part
The web did not double every year in any neat, tidy sense that search engineers could comfortably model forever. But the volume problem is real enough. AI has made content cheap, fast, and abundant, and that has already outpaced the assumptions built into a lot of publishing systems.
WordPress exposing AI connectors only inside the block editor is a small operational crack. Google having to keep adjusting its stance on AI content is the larger structural crack. Put them together and the picture is hard to miss. Platforms are still behaving as though content production is a manageable editorial process, while the internet is increasingly behaving like a machine that can produce text faster than quality systems can comfortably sort it.
That mismatch is where the next SEO fight lives. Not in whether AI can write. In whether your stack, your workflow, and the search engines in front of you can survive the volume.
