How Generative AI Could Collapse the Web’s Advertising Economy – Berners-Lee’s Warning

Tim Berners‑Lee, the inventor of the World Wide Web, warns that generative artificial intelligence is threatening the foundation of today’s web‑based economy.

Berners‑Lee notes that the unprecedented success of large language models (LLMs) is causing users to retrieve information directly from AI rather than visiting websites, which could cause the advertising‑driven internet economy to collapse.

He points out that Wikipedia’s traffic has already fallen by 8% this year as users turn to AI‑powered search engines for answers, reducing the amount of fresh data available to train future LLMs and creating a vicious cycle where AI becomes both reader and author of web content.

Inrupt, the company founded by Berners‑Lee, is developing a chatbot called Charlie that uses personal data to provide tailored responses while reminding users which platforms can access their information.

Berners‑Lee warns that if LLMs read all webpages and deliver answers without humans ever seeing the original pages, the traditional ad‑based revenue model will “immediately collapse.” Advertising relies on human eyes; AI‑generated summaries would dramatically cut click‑through rates and ad income.

He suggests that AI might serve as a long‑overdue “reset button” for the internet, especially as users grow weary of hyper‑personalized ads that feel invasive.

Berners‑Lee adds that a new model will be needed to replace the current one, though what that model looks like remains unclear.

Financial data underscores the risk: Alphabet reported $100 billion in quarterly revenue, largely driven by its cloud division, while Meta still generated $51.2 billion in Q3 revenue, both heavily dependent on advertising.

John Bruce, CEO of Inrupt, notes that major brands and payment providers are already uneasy, fearing they may soon “bow to a large language model.” Mozilla President Mark Sullman echoed the concern, describing the advertising sponsorship model as being at a “potential crossroads.”

He argues that now is the time to build systems that respect privacy and give users control, acknowledging that the existing ecosystem will not disappear overnight but will evolve rather than vanish instantly.