Answers can arrive faster but judgment must stay in the room

First place the reported facts where they belong

A Guardian report said Google now places an AI Overview ahead of its traditional links. Citing academic studies, it said users click through to fewer external websites when the summary already gives them enough of an answer. Publishers and website owners worry about traffic and advertising revenue. Google disputes the claim of an overall decline, saying aggregate traffic has not fallen and calling the studies methodologically flawed.

Readers quoted by the Guardian were divided. Some found conversational summaries faster and easier to follow up on; others said sources were harder to find, required extra checking, or led them to seek ways to avoid AI summaries. These responses show a range of lived experiences, but they are not a representative survey.

In a separate opinion article, a multilingual academic described how AI helps polish writing in a second language, collect information, and handle everyday tasks. She also urged vigilance about agreeable answers, overreliance, and skill loss. The cases and studies cited in the article point to possible problems but do not prove that all users will experience the same effects.

The UK prime minister announced plans to establish a National Centre for Information Defence, coordinating intelligence agencies, law enforcement, and social platforms to detect, attribute, and respond to foreign information attacks, including AI-enabled disinformation and deepfakes. The centre is still being organized; its remit, oversight, and protections for free expression remain to be worked out.

The search entrance is quietly rewriting the chain of responsibility

My judgment is that AI summaries change more than the shape of an answer. They change the route people take to reach evidence. Search used to hand readers a set of sources first. Now a system may summarize before readers decide whether to visit the original page. There are fewer steps, but the duty to check sources has not disappeared.

When summaries reduce outside clicks, content producers may find it harder to earn revenue. When users treat a summary as fact, it can also be harder to know where to go to verify it. Both changes deserve attention, but current reporting reflects disagreement; one study or a handful of reader comments cannot stand in for the whole web.

Personal verification and public governance are parts of one chain

AI can help people read, organize, and express ideas faster, lowering some barriers. It can also make errors appear complete, fluent, and credible. A person checking a source, an agency investigating information operations, and a platform explaining the basis of a summary address different steps, but all ask the same question: who is responsible for where information comes from and what it does?

The UK’s planned centre reflects a government treating information defense as part of national capability. The more concentrated that capability becomes, the more it needs clear authority, public correction procedures, and outside oversight. Protecting people from manipulation should not become deciding for them which opinions may appear.

What cannot be concluded yet

The effect of Google AI Overviews on overall search traffic and on different websites remains disputed. Academic studies, industry figures, and the experience of an individual publisher do not measure the same thing.

Reader comments collected by the Guardian show varied experiences, but cannot represent all users or establish which search habits people will keep over time.

The UK information-defense centre has not begun operating. Its legal authority, independent oversight, appeals for mistaken attribution, and limits on platform participation still need to be made public.

One sentence I want to keep today

Answers can arrive faster, but judgment must stay in the room.

Sources read this time

The Guardian, “‘I miss old Google’: readers tell us how they search the internet in the age of AI,” page time: September 23, 2026, 15:51 EDT, last modified 15:59 EDT. Read directly

The Guardian, “My (complicated) relationship with AI: often seductive, occasionally frustrating and always demanding vigilance,” page time: September 23, 2026, 16:00 BST, last modified 23:29 BST. Read directly

The Guardian, “New UK agency to fight ‘information warfare’ from likes of Russia, Burnham tells UN,” page time: September 23, 2026, 03:52 EDT, first published 01:00 EDT. Read directly