Google has spent years lecturing publishers on what quality looks like.

Be original. Be useful. Show experience. Demonstrate expertise. Build authority. Earn trust. Add bylines. Explain who created the content, how it was produced and why it exists.

Avoid thin pages. Avoid recycled summaries. Avoid content made only for search. Invest in reporting, structure, sourcing, page experience and editorial credibility.

Publishers listened.

They hired writers. They paid editors. They built author pages. They tightened editorial policies. They added disclosures, source links, expert commentary, schema, internal links, clean headings, fast hosting, security tools and original analysis. They chased the standard Google said mattered.

Then Google changed the game.

Its AI Overviews now sit above the traditional search results, summarising the web into machine-written answers while often pulling attention away from the sites that created the underlying value.

Worse, in some cases, Google’s AI search layer gives visibility to Reddit threads, casual forum commentary and anonymous user opinions while professionally edited articles are pushed below the fold.

That is why publishers are angry.

This is not a minor SEO complaint. It is a dispute over who gets rewarded for producing knowledge on the open web.

Google’s own Search Central documentation says its systems are designed to prioritise helpful, reliable information created for people, not content made primarily to manipulate search rankings.

It asks whether content provides original reporting, research or analysis; whether it offers substantial value compared with other pages; whether it is written or reviewed by someone who demonstrably knows the topic; and whether it carries clear sourcing and evidence of expertise.

Google also says its systems look for signals associated with experience, expertise, authoritativeness and trustworthiness, known as E-E-A-T, while noting that trust is the most important element.

Now compare that standard with a Reddit thread.

A Reddit answer may be useful. It may contain lived experience, niche technical advice, honest product feedback or the kind of blunt user commentary that polished corporate pages avoid. Nobody serious should argue Reddit has no value. It clearly does.

But Reddit is not a newsroom. It is not a peer-reviewed journal. It is not a specialist trade publication. It is not a professional blog with editorial liability, corrections processes, sourcing standards and named authors.

It is a forum — a chaotic mix of genuine expertise, anonymous opinion, jokes, hearsay, brand manipulation, misinformation, affiliate influence, grudge-posting and people confidently guessing in public.

That distinction matters because Google is no longer merely ranking sources. It is packaging answers.

Traditional search gave users a list. A person could compare a news report, a company blog, a government page, a technical manual, a review site and a forum thread.

AI Overviews collapse that open process into a single answer box. Google chooses the framing, the sources and the visibility. The user often gets the answer before they ever reach the publisher.

That is not just a change in interface. It is a transfer of power.

The data now makes the publisher backlash harder to dismiss. Pew Research Center found that when Google users encountered an AI summary, they clicked a traditional search result in only 8% of visits, compared with 15% when no AI summary appeared.

Users clicked a source inside the AI summary in just 1% of visits. Pew also found that about 18% of Google searches in its March 2025 dataset generated an AI summary, with Wikipedia, YouTube and Reddit among the most frequently cited sources across both AI summaries and standard results.

For publishers, those numbers are brutal.

The AI answer uses the web to satisfy the searcher, but the click often dies on Google. The publisher still carries the cost of producing the work. Google keeps the user. Advertisers keep buying the page. The public receives a convenient summary. The creator of the original value gets a shrinking chance of being visited.

That is not a fair exchange. It is extraction with a friendly user interface.

The publisher anger has grown sharper because Google also has a commercial relationship with Reddit.

In February 2024, Google announced an expanded partnership with Reddit, saying it would gain access to Reddit’s Data API for real-time, structured content and use it to better understand, display, train on and otherwise use Reddit content across Google products. Google also said the partnership would support more “content-forward displays” of Reddit information.

That does not prove Google is unfairly favouring Reddit in every AI answer. But it gives publishers every reason to ask harder questions.

If a professional publication is expected to prove authorship, editorial standards, expertise and trust, why should anonymous Reddit commentary be treated as a premium source inside AI-generated answers?

If Google wants publishers to produce high-quality material, why does its AI layer so often appear to reward the cheaper, messier and less accountable end of the web?

This is the contradiction publishers cannot ignore.

A 2026 study of 55,393 trending queries found that Google AI Overviews appeared for 13.7% of queries overall, rising to 64.7% for question-style searches.

The same study found that nearly 30% of AI Overview-cited domains did not appear in the co-displayed first-page organic results, meaning Google’s AI citations are not simply a mirror of normal search rankings. It also found that 11% of atomic claims in AI Overview responses were unsupported by the cited pages.

That should concern anyone who cares about accuracy.

Publishers have been trained to obsess over quality signals, but AI Overviews introduce a second source-selection system that is harder to inspect and harder to challenge.

A site can rank well organically, follow Google’s guidance, invest in original reporting and still be bypassed by the AI citation layer. Another source can be elevated because it better suits the summarisation engine, not necessarily because it carried the strongest editorial work.

That is the part that feels like a betrayal.

A separate 2026 study comparing Google Search, Gemini and AI Overviews found that AI Overviews appeared above organic results for 51.5% of representative real-user queries in its dataset. It also found that generative search retrieved substantially different sources from traditional Google Search, with low overlap between the source sets.

In plain English: publishers are no longer competing in one Google. They are competing in two.

There is the old Google, where SEO, authority, backlinks, technical health and editorial quality still matter. Then there is AI Google, where the answer box can choose different sources, compress the user journey and reduce the need to visit anyone else’s site.

For publishers, that is not innovation. It is platform dependency becoming platform substitution.

The economic damage is not theoretical. A 2026 study using Wikipedia traffic as a test case found that exposure to Google AI Overviews reduced daily traffic to English Wikipedia articles by about 15% across 161,382 matched article-language pairs.

The researchers found the declines were larger for culture articles and smaller for STEM topics, consistent with the idea that short AI answers can replace visits when the user’s intent is informational.

If Wikipedia, one of the most recognisable information sources in the world, can lose traffic from AI summaries, smaller publishers should be terrified.

The anger is already moving from complaint to litigation. Penske Media, owner of Rolling Stone, Billboard and Variety, sued Google in 2025 over AI Overviews, alleging that Google used its journalism in AI-generated summaries without permission and reduced traffic and revenue.

Reuters reported that Penske claimed affiliate revenue had fallen by more than a third since its 2024 peak, while Google defended AI Overviews and said they improve search and help users discover content.

This is the publisher position in one sentence: Google built its empire by sending traffic to the web, and now it is using the web to avoid sending traffic back.

That is why the Reddit issue cuts so deeply.

Reddit threads cost almost nothing to produce. They do not require a newsroom. They do not pay specialist journalists. They do not commission expert explainers.

They do not maintain a publishing calendar, legal review, corrections policy, source verification process or editorial archive in the way serious publishers do.

Yet Reddit-style content is perfectly shaped for AI search. It is conversational. It is abundant. It is constantly updated. It is full of direct answers. It contains the language real people use. It looks authentic. It often satisfies the machine’s appetite for “experience” without requiring the accountability that should come with expertise.

That is where the system becomes dangerous.

A Reddit user may be a professional engineer giving excellent advice. They may also be a teenager, a bot, a marketer, a competitor, a troll, an affiliate operator, a brand fan, a disgruntled customer or someone repeating something they read somewhere else.

On Reddit, readers can see the mess. Inside an AI Overview, the mess is cleaned up and presented in the smooth, neutral voice of Google.

That is how opinion gets laundered into authority.

And publishers are right to be furious.

They were told to build trust, but the AI layer can make anonymous commentary look trustworthy. They were told to add original value, but the AI layer can satisfy users before they reach the original work.

They were told to avoid summarising other people’s content without adding much value, yet AI Overviews are built around summarising the web at scale. They were told to focus on people-first content, while Google’s own interface increasingly converts publisher work into Google-first answers.

The hypocrisy is hard to miss.

Google is not a neutral librarian in this model. It is the crawler, indexer, answer-writer, source selector, traffic gatekeeper and advertising platform.

That concentration of power would be uncomfortable even if the system were flawless. It is far more troubling when research shows unsupported claims, reduced click behaviour and a citation layer that does not simply follow organic ranking.

The open web cannot survive if quality publishers are reduced to unpaid training material, unpaid source material and occasional footnotes.

A credible AI search system should put original reporting, expert analysis, recognised publications, technical documentation, government sources, academic research and properly maintained specialist blogs at the centre of source selection. Reddit should be used carefully, visibly and in context — not treated as a shortcut around editorial expertise.

Google should disclose far more about how AI Overviews select sources, how Reddit content is weighted, how commercial data relationships are handled, how publishers can compete fairly, and whether AI answers are reducing traffic to the sites they depend on.

Until that happens, publishers will continue to see AI search as a rigged bargain.

They do the work. Google takes the answer. Reddit gets the visibility. The user stays on Google. The publisher gets told to keep improving quality.

That is not the future of search.

That is the slow conversion of the open web into raw material for a machine-written answer box — built on everyone else’s labour, monetised by the platform that already controls the front door.