The open web runs on a simple bargain: creators publish content, search engines and social platforms send them readers, and advertising or subscriptions pay the bills. Artificial intelligence is now rewriting that bargain — and, according to internal Microsoft documents reported this week, people inside one of the world’s biggest technology companies have worried that AI could push the web into a damaging “doom loop.”
The concern, as reported by The Cooldown, is straightforward. If AI assistants answer questions directly, fewer people click through to the websites that produced the original reporting, photography, reviews and expertise. With less traffic, publishers earn less money. With less money, they produce less original content. And with less original content, the AI systems that depend on a rich, constantly updated web have less good material to learn from and summarise. Each turn of the loop makes the next one worse.
How the Loop Would Work
Start with a recipe, a product review or a local news story. Creating it costs time and money. Traditionally, a search referral rewarded that effort with an audience. An AI summary changes the exchange: the user gets the answer, the platform keeps the attention, and the creator may get little more than a small citation link that few people follow.
Multiply that across millions of queries and the economics of publishing shift. Large publishers can negotiate licensing deals with AI companies, and some have. Smaller sites — specialists, local outlets, independent reviewers — rarely have that leverage. Our earlier reporting, how AI agents are quietly taking over everyday office work, shows how dependent smaller technology publishers are on steady referral traffic to survive.
Why Microsoft’s Warning Matters
Microsoft is not an outside critic of AI search. It operates Bing, builds Copilot and has invested heavily in OpenAI. A warning emerging from inside such a company therefore carries particular weight: it suggests the tension is recognised even by the firms building the products. The same companies need a healthy web twice over — as a source of training and retrieval material, and as an ecosystem of partners whose content makes their assistants useful.
There is also a quality risk. If the open web fills up with low-cost, AI-generated pages produced mainly to be summarised, while original reporting shrinks, the information supply itself degrades. Models can begin learning from synthetic material that echoes other synthetic material, errors included. Researchers sometimes call this a feedback problem; the reported “doom loop” phrase captures its economic twin.
Reasons Not to Panic — Yet
The web has survived predictions of its death before. Social media was supposed to end publishing; instead, publishers adapted with newsletters, podcasts, video, memberships and carefully planned digital campaigns built around content, design and reach. AI could also create new value for creators: licensed datasets, referral features in chatbots, agent-driven commerce and tools that let small teams produce better work. Some AI products already experiment with prominent citations and publisher partnerships, and regulators are examining how platforms share value, as the BBC’s technology coverage regularly documents.
Much depends on design choices made now. If assistants make it easy — and worthwhile — for users to visit sources, the loop can be broken. Clear attribution, real click-through data shared with publishers, fair licensing and limits on scraping for those who opt out would all help align AI’s growth with the web that feeds it.
The Web AI Needs Is the Web It Must Not Starve
The deeper point of the warning is not anti-AI. Assistants that help people find, compare and understand information are genuinely useful, and readers benefit from them every day. But usefulness built on other people’s unpaid work is not a stable foundation, either commercially or culturally.
Publishers, platforms and AI companies are now negotiating, in real time, what the next version of the web’s bargain looks like. Getting it right matters far beyond media businesses. A web with fewer original voices is a poorer source of knowledge for everyone — including the machines. The doom loop is a warning, not a destiny, and the choices that decide which it becomes are being made this year.
