Google pulled a subset of its AI Overviews after users received dangerous health guidance, The Guardian reported. The publication's headline characterized the pullback as "dangerous and alarming." The move targets generated summaries that produced health answers deemed unsafe for users.

What does the source headline confirm?

Two facts sit inside the Guardian headline. First, Google removed some AI summaries. Second, the action followed reports that users' health was put at risk.

The headline does not specify a date, geography, or number of removed summaries. Those datapoints will likely appear in the Guardian's full article or in Google's own public statement on the change.

How far has the AI Overviews rollout reached?

Google launched AI Overviews in May 2024, initially for logged-in U.S. users. The feature placed AI-generated summaries above traditional blue-link results. Google extended the rollout to the U.K., India, and additional markets through 2024 and into 2025.

Industry tooling vendors have tracked AI Overviews volatility since the public launch. The feature surfaces cited publishers inconsistently, with monitoring tools showing weekly inclusion changes for top health domains. Health and finance queries have produced the highest volume of public complaints across those tools, per third-party tracking platforms.

Why are health queries the first target?

In May 2024, AI Overviews produced a widely shared answer suggesting users eat a small amount of glue to fix pizza cheese. That output circulated broadly through social platforms. Google responded at the time by calling the query uncommon and the feature broadly functional.

The current pullback suggests a faster response than in 2024. The Guardian's headline language — "dangerous and alarming" — sits sharper than 2024 coverage, which leaned on softer terms like "uncommon errors." Google's YMYL — "Your Money or Your Life" — policy bucket also pushes health queries into the strictest accuracy bucket inside the quality rater guidelines.

How does the pullback split by site type?

The pullback splits by vertical. Health publishers lose direct AI Overview citations and the referral traffic those citations produced. General news publishers face smaller losses across the board. Local services and e-commerce sites appear unaffected by this specific reported incident.

Publishers running FAQ-style health pages face the steepest near-term drop. Those pages surfaced as AI Overview sources more often than editorial long-form articles during 2024 testing, per widely reported industry analysis.

What should publishers and SEOs monitor next?

Three signals will measure the seriousness of the pullback over the next reporting cycle:

• The full Guardian article, which should name the on-record source behind the "dangerous and alarming" characterization.
• A Search Central blog post or Search Liaison social statement confirming scope, geography, and timing.
• The recovery curve — whether removed queries stay free of AI Overviews or return after new guardrails.

Google has historically posted AI Overview changes through Search Central for major adjustments. That channel will carry the authoritative version of this episode. Watch the next publication date there, not Google's marketing-side announcements, for the verified scope.