OpenAI and Microsoft Knew They Were Creating a 'Doom Loop' for the Web

The 'Doom Loop' for the Web

OpenAI and Microsoft understood the mechanics of what they were setting in motion. The web depends on a simple bargain: publishers create content, and search engines send traffic back in return. AI chatbots break that bargain by answering questions directly, so users never click through to the original source.

Fewer visitors mean less advertising revenue for publishers. Less revenue means fewer resources to produce the reporting, guides, and analysis that AI models are trained on in the first place. As the quality of freely available content declines, the models themselves have less fresh material to learn from.

This is the doom loop: each turn of the cycle starves the ecosystem that feeds it. According to reporting by The Atlantic, both companies were aware of this dynamic. They knew that widespread AI adoption would drain the traffic that sustains the open web, and they moved forward regardless.

Driving Toward Google Zero

This was not an accident. As the blog Daring Fireball observed in a post titled "The Google Zero Era Is Here," the people running these platforms understood exactly where the road led. Google itself popularised the term "Google Zero" to describe a future in which the search engine answers questions directly, so users never need to click through to the open web at all.

That future did not arrive by itself. It was built, deliberately and step by step, by companies that could see the destination clearly. Google trained users to expect instant answers on the results page. AI companies then trained them to expect a single generated response instead of a list of sources.

Each decision made sense in isolation. Together, they formed a pattern: the companies knew they were driving us toward Google Zero, and they kept their foot on the accelerator anyway. The incentives pointed that way, and nobody with the power to change course had a reason to.

They Did It Anyway

OpenAI and Microsoft understood the risk. Both companies could see that scraping the open web at scale would damage the very ecosystem their models depended on. They knew publishers were already struggling, and they knew that flooding search results with AI-generated summaries would strip sites of the traffic that funds them. None of this was a mystery.

They proceeded regardless.

The incentives were simply too strong to resist. Whoever trained the largest model on the most data would win the race, and the race was framed as existential. Slowing down to negotiate licensing deals or respect robots.txt meant handing an advantage to a competitor. So the crawl continued, the summaries shipped, and the consequences were treated as someone else's problem.

This is the pattern that defines the current moment. The harm was foreseeable, publicly discussed, and repeatedly warned about by the people building the tools. And still, the decision was made to move forward, because the cost of restraint was immediate and the cost of damage was distant.

The Impact on the Web

The consequences of this doom loop extend far beyond a single company's fortunes. As Google's AI Overviews answer more queries without a click, and as publishers respond by blocking crawlers, the open web that both sides depend on begins to erode. The article frames Google Zero not as a distant hypothetical but as a destination the industry is actively driving toward, one decision at a time.

Publishers who once traded free access for traffic now find that bargain collapsing. Blocking AI crawlers protects content in the short term but removes it from the very systems that surface it. Meanwhile, users lose the diverse range of sources that made the web worth searching in the first place.

The irony is sharp: the companies building AI products were trained on the web's collective output, and now their success threatens the incentive to keep producing it. If the doom loop continues, the web could become a smaller, poorer place — one where the answers are plentiful but the sources behind them have quietly disappeared.