Google's AI Answers Are Draining the Open Web

For twenty-five years, a Google search returned a list of blue links and sent you off to read the pages behind them. That bargain is breaking. Google now answers a growing share of queries directly, with summaries written by its Gemini models, and the sites that supplied the underlying information are watching their traffic drain away.

The shift has a name inside Google: AI Mode, unveiled at the company's I/O conference this year and powered by the fast Gemini 3.5 Flash model. Instead of ten ranked results, it composes an answer and folds a few source links into the text. Google has called it the most significant change to Search in more than two decades. Publishers have a blunter description.

The numbers behind the panic

Zero-click searches, where the user gets what they need without visiting any website, now make up around 60 percent of queries. A study by the analytics firm Ahrefs, published in February, found that pages showing an AI Overview lost about 58 percent of their click-through rate at the top of results, nearly double the drop measured a year earlier. Press Gazette reported that Google search traffic to publishers fell 33 percent globally in the year to November 2025, and 38 percent for publishers in the United States.

Individual sites tell the same story in sharper detail. HubSpot estimates it lost 70 to 80 percent of its organic traffic. The education company Chegg reported a 49 percent decline and has cited AI answers in an antitrust complaint. DMG Media, publisher of the Daily Mail, documented drops as steep as 89 percent on some queries. NPR reportedly called the trend an extinction-level event for online news.

Google's case, and the catch

Google argues the picture is more nuanced. It says AI Mode surfaces a wider range of sites, sends better-qualified visitors, and now adds further-reading links to push people outward. There is something to that. A reader who works through a summary and still clicks arrives more interested than someone who bounced off a headline.

The catch is circular, and it is the uncomfortable part of the story. AI answers are built on the work of the same publishers whose economics they are undercutting. If the open web stops paying for reporters, researchers and reviewers, the models lose the fresh, reliable material they summarise in the first place. Google is not the only pressure here either. Sites are already trying to force AI crawlers to pay for access, and the shopping and video fronts are moving too, as Google's own Gemini Omni Flash shows.

None of this means the AI answers should not exist, or that they are always wrong on the facts. But convenience has a way of hardening into trust, and trust in a machine-written summary is worth examining before you lean on it. We take that question up separately in the myth that an AI search answer is always right.