For years, businesses were told to play by Google’s rules.

Publish useful content, answer customer questions, build authority, earn links, keep your website fast, make it trustworthy.

Do the work, and Google would reward you with visibility.

That bargain is now breaking down.

Across the internet, publishers, retailers, local businesses, review sites, bloggers, trades, comparison platforms and specialist information websites are watching search traffic shrink.

In many cases, the loss is not abstract. It shows up as fewer phone calls, fewer bookings, fewer ad impressions, fewer subscribers, fewer sales and fewer reasons to keep producing the content Google has long depended on.

Google did not build the open web. But it became the main road through it. Now it is placing its own AI answer machine at the front gate and asking everyone else to survive on whatever traffic leaks through.

The company presents AI search as progress. A faster answer. A smarter interface. Less friction for the user. But the public should be more sceptical.

What Google calls convenience may also be one of the largest transfers of value in the history of the internet: from the businesses and publishers that create information to the platform that summarises it.

The old search model was far from perfect. It was cluttered with ads, shaped by SEO games and too often dominated by large sites.

But at least it still worked on a basic exchange. A user searched. Google showed links. The user chose where to go. The website received a visitor. That visit could become revenue.

AI search changes the deal.

Instead of directing people to the source, Google increasingly gives them a polished summary above the links. The answer appears complete enough that many users do not need to click.

The original website may have paid writers, editors, photographers, analysts, researchers, web developers and support staff to create that information. Google’s AI can then compress it into a few paragraphs and keep the user on Google.

That is not just a product update. It is a business model shift.

For a news publisher, fewer clicks mean fewer ad impressions and fewer chances to convert readers into subscribers. For a local business, fewer visits can mean fewer quote requests.

For a recipe site, fewer page views can mean less income from the ads that pay for testing, photography and writing.

For a niche expert site, it can mean years of work reduced to an unattributed sentence in a box.

Google may argue that AI summaries still include links. Technically, they often do. But a link placed beneath an answer is not the same as a search result that sends the reader outward.

When the answer is already served, the incentive to visit the source is weakened. That is the point. The friction has been removed — and with it, the traffic.

The bigger problem is that AI answers are not necessarily better answers.

Traditional search made users compare sources. It showed a spread of results. The reader could open a government page, a news article, a business website, a forum thread or an expert analysis and decide which one deserved trust.

That process was messy, but it gave the public some visibility over where information came from.

AI search does something different. It blends material into a single confident response. It may look clean, but clean is not the same as correct.

Unique answers don’t not mean accurate

AI systems are good at producing plausible language. They are not inherently good at knowing what is true.

They can surface a unique answer, but unique does not mean accurate. They can sound certain while missing context, flattening nuance or relying on sources the user never sees.

That matters because search is not just a consumer tool. It is public infrastructure.

People use Google to make decisions about health, money, law, politics, travel, repairs, products, schools, housing and local services.

When the search engine becomes the answer engine, Google is no longer merely organising information. It is interpreting it. It is choosing what matters, what gets compressed, what gets omitted and what gets visibility.

That is editorial power, whether Google wants to call it that or not.

The public interest issue is not only that businesses are losing traffic. It is that the incentive to create reliable information is being damaged.

If websites cannot earn revenue from the work they publish, fewer will invest in that work.

The web will not collapse overnight. It will thin out.

Specialist writers will disappear. Smaller publishers will shut. Local businesses will stop maintaining detailed guides.

Independent reviewers will give up. The material AI depends on will become weaker, cheaper and more recycled.

Then Google’s AI will be left summarising a web it helped hollow out.

Supporters of AI search will say users prefer quick answers. Of course they do. People also prefer free journalism, free maps, free videos and free expert advice.

But “free” usually means someone else is paying. Online, that someone has often been the publisher, the small business or the creator who accepts the cost because traffic can turn into income.

Remove the traffic and the economics fail.

This is why regulators are beginning to pay attention. The question is no longer whether Google has improved search for users in the short term.

The question is whether it has used its dominance to rewrite the rules for everyone else.

When one company controls the main doorway to the internet, it should not be allowed to take content from the rooms beyond that doorway, summarise it, monetise the attention and leave the original creators with scraps.

Google’s AI search was not necessary in the way the company suggests. Search did not need to become a machine that answers everything. It needed to become cleaner, more transparent and less polluted by spam.

It needed better source labelling, fewer low-quality results and more respect for original reporting and expertise. The answer to a damaged search experience was not to bury the web under an AI layer.

The old model worked better because it preserved a basic democratic function: it showed people where information lived. It let them move across the web. It kept some power with the reader and some revenue with the creator.

AI search centralises both.

There is still time to fix this, but only if Google is forced to treat the web as an ecosystem rather than a supply chain.

Publishers and businesses should have the right to opt out of AI summaries without disappearing from normal search. They should receive clear attribution when their work is used.

There should be transparent reporting on how AI results affect traffic. And where Google’s products rely on commercial content, there should be compensation.

This is not nostalgia for an older internet. It is a warning about the next one.

A search engine that answers every question by absorbing everyone else’s work may feel useful today.

But if it destroys the businesses that produce that work, the public will be left with something poorer: fewer sources, fewer voices, less accountability and a web where the most powerful company in search decides not just what we find, but what we know.

Google has not simply changed search.

It has changed who gets paid for being cited. And not necessarily for providing the right answer