Two things happened to the web this past year that were reported as separate stories. The first: Google finished dismantling the old extension system that made ad blockers powerful, and the last workarounds for running the full uBlock Origin in Chrome are nearly gone. The second: Google’s AI-generated answers got good enough that people stopped clicking through to the sites those answers were summarized from, and by 2026 fewer than a third of Google searches still sent a click to the open web. Treated apart, each is a tidy tech-policy item. Put side by side, they describe something bigger: the terms of the open web are being quietly rewritten, and mostly by one company acting on its own incentives.
What actually changed in your browser
The technical shift sounds boring and is not. Chrome’s move from the old extension architecture (Manifest V2) to the new one (Manifest V3) replaced the mechanism ad blockers used to inspect and filter web requests in real time with a more limited, pre-declared system. The practical result is measurable: independent testing suggests the original uBlock Origin blocked ads at a 90-to-95 percent rate, while the Manifest V3-compliant “Lite” version drops to roughly 60 to 70 percent. The full-power tool did not get worse; it stopped being allowed to run in the world’s dominant browser.
Google’s stated case is security and performance — the old API was powerful in ways that could be abused, and a more constrained system is easier to keep safe. That argument is not baseless. But it collides with an inconvenient counterpoint from the government’s own security experts: the U.S. Cybersecurity and Infrastructure Security Agency has recommended ad-blocking software as a defense against “malvertising” — malware distributed through legitimate ad networks. When your national cyber-defense agency treats ad blockers as a security layer and the browser vendor treats them as a security risk, it is worth asking whose security each party has in mind.
The part that’s hard to explain away
Here is the fact that makes the whole thing hard to read as pure coincidence. The same company that constrained ad blockers in its browser earns the overwhelming majority of its money from ads — an estimated $239.5 billion in advertising revenue in 2025. I want to be careful here, because motive is not the same as proof, and a company can make a defensible engineering decision that also happens to serve its bottom line. But the burden of persuasion sits with the party whose safety argument aligns so neatly with its revenue. Chrome did not ban ad blocking. It capped the version of ad blocking that worked best against the exact ad systems its parent company profits from. Reasonable people can call that prudent platform hygiene; reasonable people can also call it something less flattering. The architecture does not tell you which, and that ambiguity is the point.
Meanwhile, the sites themselves are starving
The second shift is larger and less reversible. For twenty years the open web ran on an implicit bargain: Google indexed your content and, in exchange, sent you readers you could show ads to or convert into subscribers. AI-generated answers break that bargain by delivering the substance of your page directly on the results screen, so the user never arrives. The data is no longer speculative. Press Gazette’s analysis found global Google search referral traffic to publishers fell roughly 33 percent in the year to November 2025, with some major news sites losing between a quarter and half of their search-driven traffic.
The single most clarifying data point in the whole shift involves Wikipedia. It is the most-cited source in Google’s AI answers, and it still saw human pageviews fall about 8 percent. Sit with that. Being the most quoted reference on the internet no longer reliably brings visitors, because citation and traffic have been decoupled. The answer takes your work; the footnote is your consolation prize. For a volunteer encyclopedia that runs on donations from readers who show up, “cited everywhere, visited less” is not a footnote — it is an existential math problem, and it applies to every site that lives on referral traffic.
Why these two stories are really one
Line them up and the shape is clear. One change reduces your ability, as a reader, to control what the web does to you. The other reduces the ability of the sites you read to survive the web being done to them. Both run through the same chokepoint: a single company that operates the dominant browser, the dominant search engine, and one of the largest ad businesses on earth, and that is now also the gatekeeper deciding which answers you see without leaving its property. None of the individual moves requires a conspiracy. They require only that a company with that much surface area keep making locally rational choices, each of which happens to route more attention and money through itself.
The uncomfortable conclusion is that the open web’s health increasingly depends on the restraint of an entity with little incentive to restrain itself — and that the alternatives are real but effortful. Firefox still runs the full-strength blockers because it is not built on Chrome’s framework. Some publishers are surviving by building direct relationships — newsletters, subscriptions, communities — that do not route through search at all. Both are acts of deliberate independence from the default, and defaults are exactly what most people never change. That, in the end, is the quiet lever: the open web was built on the assumption that no single party controlled the road, the toll booth, and the destination at once. That assumption is now the thing being tested, and the test is not going especially well.
