R.I.P. World Wide Web, how AI is becoming a killer

R.I.P. World Wide Web, how AI is becoming a killer

AI is reshaping the web faster than any past disruption. Two dynamics stand out. On one side, platforms keep users inside their walls with zero-click results and AI overviews. On the other hand, the internet is filled with synthetic “slop” that makes trust and originality harder to sustain. Together, these shifts threaten the economic engine of the open web (+ what it means for e-commerce).

What you’ll learn in this edition

  • Why the zero-click internet is becoming a reality
  • How AI-Generated “slop” is flooding the web
  • Why the “Dead Internet” theory resurfaces
  • How traffic plunges and creators are impacted
  • What the responses to these developments are
  • How all this impacts e-commerce as we know it

The broken bargain

The web once relied on a simple exchange: publish content, get traffic, monetise that traffic. AI is breaking the chain.

Cloudflare CEO Matthew Prince warns that generative AI and new search paradigms are killing the web’s business model by removing the need to visit websites at all. He points out that Google used to send one visitor for every two pages it crawled. Today it can take six. Large models like OpenAI’s scrape thousands of pages per prompt and send almost nothing back.

You can find more on the topic and see what people on Reddit think in the link below:

In fact, about 75% of Google queries now get answered without the user leaving Google. Google’s AI Overviews show the shift clearly. Summaries answer user intent at the top of results, often with images taken from blogs, but without a click. Pew research backs this, and Wired has shown how such features cannibalise clicks that once sustained media outlets. Social media platforms similarly discourage external clicks, favoring content hosted internally.

Slop and “enshittification”

Beyond just search, another way AI may be “killing” the internet is by flooding it with low-quality, automatically generated content, sometimes referred to as “AI slop.” This term “slop” denotes the next evolution of spam: text, images, videos generated en masse by AI with little human curation or value. Guardian columnist Arwa Mahdawi warns that the internet is rapidly being overtaken by AI slop,” making genuine human-created content harder to find.

  • A recent study estimates that more than half of longer LinkedIn posts may already be AI-written.
  • Newsrooms have tested AI-written stories under fake author names.
  • Social feeds are full of uncanny AI-generated images, like the infamous “Shrimp Jesus” meme on Facebook.

Article content
The infamous "Shrimp Jesus" as an example of "entshittification"

Cory Doctorow calls this “enshittification”: platforms optimising for clicks at the expense of quality. Nieman Journalism Lab reports that AI-driven spam accounts even gained reach thanks to Facebook’s algorithms, while MIT Tech Review warns of feedback loops where AI trains on AI-generated junk, further degrading quality.

Low-quality ‘slop’ generated by AI is crowding out genuine humans across the internet, but instead of regulating it, platforms… are positively encouraging it. Where does this end?

Arwa Mahdawi, Guardian columnist and author

The “dead internet” theory

Amid these trends, an old idea has gained new relevance: the Dead Internet Theory. This conspiracy theory posits that a majority of online activity and content is already generated by bots or AI, not people, effectively that the internet “died” sometime around 2016 and is now mostly fake. According to the theory’s proponents, government or corporate actors might be behind masses of social bots and algorithmically generated posts, seeking to manipulate public opinion or simply replace messy human discourse with something easier to control.

While the more grandiose claims are not taken seriously by experts (there’s little evidence of a centrally orchestrated plot to supplant all humans online), the observable facts about bot traffic are real.

Have you ever asked yourself what the chances are that you are interacting with an AI at any given moment on the Internet? It turns out, it’s quite high.

Tech commentator Ignacio de Gregorio Noblejas

Many internet users have indeed noticed that online reviews, forum posts, or even news articles often have a formulaic, machine-written feel. As generative AI tools proliferate, our baseline assumption may shift toward skepticism: assuming content might be AI-made unless proven human. In de Gregorio’s words, some fear “the Internet as we know it is dying.”

Enrique Dans counters that the web is not dying, but evolving. He expects reinvention rather than extinction: niche communities and investigative voices still matter, regulation or licensing may emerge, and an “agentic web” of standards-based bots and microtransactions could realign incentives while preserving diversity.

Both views highlight the same tension: how to preserve authenticity in a machine-saturated space.

Impact on websites and creators

One immediate, measurable effect of these AI disruptions is a sharp decline in web traffic for many sites. As more queries are answered by AI or kept within walled gardens, fewer clicks reach the independent web.

Article content
From Similarweb report "The Impact of Generative AI: Publishers

The pain is visible. A brief overview of some traffic/click declines:

  • MailOnline (Daily Mail): –56% CTR when AI Overviews appear. Even holding the #1 organic position, MailOnline reported clicks drop by 56% on desktop (48% on mobile) whenever an AI Overview sits above the results.
  • Similarweb (sector-wide): –26% to news publishers; zero-click up to 69%. Since AI Overviews’ May 2024 launch, Similarweb reports –26% organic traffic to news sites and zero-click searches rising from 56% → ~69% YoY (May 2025).
  • Penske Media (Rolling Stone, Billboard, Variety): lawsuit alleging declines >⅓. In Sept 2025, Penske sued Google, claiming AI Overviews repurpose its journalism and have driven >33% declines from 2024 peaks.
  • Dangerous-Business (Blog): Travel blogger Amanda Williams lost 40% of her traffic and 34% of ad revenue in a year.

Google initially pushed back on the idea that its AI features were harming the open web. In May 2025, Google’s VP of Search, Nick Fox, insisted “from our point of view, the web is thriving”, and Google’s Search team claimed that outbound click volume remained “relatively stable” year-over-year despite AI changes.

However, a few months later, Google seemed to contradict itself in a legal context. In a September 2025 court filing (part of an antitrust case), Google acknowledged “the open web is already in rapid decline,” arguing that breaking up Google’s ad business could “accelerate that decline”. This stark admission, highlighted by journalist Will Oremus as Google “can’t decide if the web is thriving or dying”, underscores what many publishers have been saying all along: the open web’s health is indeed deteriorating in the age of AI.

Pushback and experiments

Responses are emerging, though fragmented.

  • Bot tolls and pay-per-crawl: Cloudflare now lets sites block or charge AI scrapers.
  • Licensing deals: News Corp struck agreements with OpenAI; Reddit, Inc. charges Google $60m/year for API access.
  • Policy: The EU’s draft AI Act pushes for transparency on training data.
  • Detection tools: Imperva ’s research shows bots already account for nearly half of web traffic, making provenance tracking essential.

Underlying all these responses is a recognition that the open web is at a crossroads. As Prince said, this is an “existential threat” even to companies like Cloudflare: “If the internet stops existing, what’s left for Cloudflare to do?. The broader community needs to find a balance where AI can progress and provide value without destroying the incentive to create the very content it relies on. That might mean adjusting legal definitions of fair use, developing metadata that credits sources in AI outputs, or using watermarking and detection to identify AI-made material so that human work can be distinguished (and perhaps privileged). Thus far, no one solution has emerged as a silver bullet, it will likely require a combination of approaches and ongoing adaptation.

What this means for the future of e-commerce

If AI is reshaping how people search, discover, and trust content, e-commerce will inevitably feel the impact. The stakes are high: product discovery, traffic, and conversion flows are built on the very web dynamics now under pressure.

E-commerce companies are already feeling the squeeze from lower traffic volumes as AI reshapes product search. As generative AI platforms increasingly act as product discovery engines (answering queries directly within chat interfaces) fewer consumers reach online stores through traditional search. This shift means less pre-purchase browsing on brand websites and a growing need for merchants to optimise visibility inside AI-driven ecosystems.

Three big shifts stand out:

  • From SEO to AIO/GEO (Artificial Intelligence Optimisation/ Generative Engine Optimisation): Just as brands once optimised product pages for Google, the next frontier will be optimising for AI assistants. The question will no longer be “How do I rank on page one of search results?” but “How do I get my product into an AI’s shortlist?” Retailers may need to rethink feed management and product data so their offers surface in conversational AI recommendations (Conversational & Agentic Commerce). Lengow, for example, allows merchants to appear in ChatGPT's shopping experience.
  • Zero-click shopping journeys: If users get answers inside AI chats, fewer will land on brand websites. This could redirect discovery flows toward walled AI ecosystems, with e-commerce brands forced to negotiate visibility within them. Amazon links already dominate many AI answers, which raises the risk of further platform dependence. But it also opens opportunities for niche players if AI tools diversify their sources.
  • Trust and authenticity as differentiators: In a flood of AI-generated “slop”, consumers may gravitate more strongly toward brands that convey authenticity, verified information, and unique value. Rich product data, transparent reviews, and direct-to-consumer experiences will matter more. The challenge for e-commerce marketers will be to stand out against a backdrop of machine-made content.

Some experts argue this could be a forcing function: pushing e-commerce away from a race-to-the-bottom in ad spend and SEO hacks, and toward genuine product quality, stronger communities, and direct relationships with customers (via newsletters, loyalty programmes, private sales).

At the same time, new business models may emerge. Cloudflare’s pay-per-crawl vision suggests a future where AI assistants micro-pay retailers for product data each time they fetch it. Imagine your catalogue not just powering Google Shopping ads, but also fuelling AI agents that compensate you directly for every recommendation served.

In short: the death of the “old web” could be the birth of new e-commerce channels. Brands that invest in clean, structured, enriched product data and that experiment early with AI-driven discovery tools may find themselves not just surviving this shift, but thriving in it.

Closing thought

AI is undermining the open web’s vitality, but this is not its obituary. If systems emerge that credit and compensate human work, the internet will evolve rather than expire. If not, we risk a loop of machines feeding machines and a poorer web for everyone.

As The Economist put it, the open web faces an “existential threat” but its story isn’t over yet.

Adrian Gmelch

Head of Content & PR @ Lengow | Communication, Events & Media Relations Specialist | SaaS, B2B, Tech, eCommerce ✅

1w

My humble opinion: I believe that in times of AI-generated content, one thing matters above all: new, original, and value-adding content. In essence, GenAI means that we content creators have to produce even better content than before! Quality and relevance above all else, quite a demanding challenge btw! 😅 🙃

To view or add a comment, sign in

More articles by Lengow

Others also viewed

Explore content categories