How to Turn Off AI Overviews in Google (2026 Guide)

Why people search for “how to turn off AI Overviews”

Google AI Overviews rolled out widely across 2024 and 2025. For some people they’re useful — a short summary at the top of the results saves time. For others they’re frustrating: the panel takes up the top half of the SERP, the AI answers some queries inaccurately, and you have to scroll past it to reach the classic blue links that used to be right there.

In Google Trends, defensive searches like “turn off AI Overviews” and “how to disable AI Overviews” show up among the rising queries alongside “AI Overviews.” That’s an indication of interest, not a representative measure of demand — rising queries tell you the direction of growth, not the volume or share.

This guide covers the main methods people use today to limit or bypass AI Overviews — from account settings to web-only search and alternative search engines. As a bonus: how to limit the way your own site’s content is used in AI Overviews (for publishers who see a drop in organic CTR on informational, non-brand queries).

The short answer (as of May 2026):

• Google doesn’t offer a global off switch for AI Overviews.
• Three routes actually work for users: the Search Labs experiment (account-dependent), the &udm=14 URL parameter (web-only mode), or switching to Brave Search / DuckDuckGo.
• For your own site, you can only limit how your content can be used (data-nosnippet, nosnippet, max-snippet:0) — not “turn off the AI Overviews crawler.” Deploy it selectively after segmenting your queries, not site-wide.

Method 1 — Search Labs (experiment in your Google account)

The easiest method for signed-in users:

1. Sign in to your Google account at google.com
2. Open Search Labs: https://labs.google.com/search or via the profile icon → Settings → Search Labs
3. Find the “AI Overviews and more in Search” experiment
4. Flip the toggle to Off

Typically works on: Chrome, Edge, Firefox, Safari (signed-in Google account, desktop).

The catch: if you see the “AI Overviews and more in Search” experiment in your account, turning it off can limit experimental AI features. But it isn’t a guaranteed global switch for all AI Overviews — Google may still show them outside Labs, and the availability and effect of the setting depend on your account, region, and current rollout state. If you don’t see the Search Labs experiment at all, this method won’t work for you.

Method 2 — &udm=14 web filter (URL parameter)

This parameter switches the results into web mode, where AI Overviews and some other expanded SERP features (People Also Ask, featured snippets) typically don’t appear — it returns just the classic blue links.

⚠️ Important: &udm=14 isn’t an official long-term directive. It’s a web filter whose behavior Google can change at any time. Some SERP elements may differ by country and query. It isn’t a guaranteed permanent off switch.

How to use it once

After searching, add this to the end of the URL:

https://www.google.com/search?q=your+query&udm=14

How to set it as the default

Add a custom search engine in your browser:

Chrome / Edge / Brave:

1. Settings → Search engine → Manage search engines
2. Add new search engine:
   • Name: Google (web only)
   • Shortcut: gw
   • URL: https://www.google.com/search?q=%s&udm=14
3. Set it as the default

Firefox:

1. Right-click the Google search bar → Add a Search Engine
2. Use the URL above
3. Set it as the default under Settings → Search

Once it’s on, it typically returns classic blue links with no AI Overviews panel — until Google changes how the parameter behaves.

Method 3 — Switching your default search engine

⚠️ Don’t confuse the browser with the search engine. Brave and DuckDuckGo offer both a browser and their own search engine. If you install the Brave browser but leave Google as your default search engine, you’ll still see AI Overviews — the Brave browser uses whatever search engine you set. AI Overviews are a function of the search engine, not the browser.

Brave Search

Brave Search is Brave’s own search engine. It doesn’t implement the Google AI Overviews panel. Brave Search has its own AI Answer feature that’s opt-in (not the default) — if you want search with no AI features at all, check the settings.

DuckDuckGo

DuckDuckGo is a long-standing privacy-first search engine. It doesn’t implement Google AI Overviews. It has its own DuckAssist (AI summaries on some Wikipedia answers) — again opt-in.

Ecosia / Startpage / Mojeek

More privacy-first alternatives. None of them implement the Google AI Overviews panel. They may have their own AI/answer features, though — check the settings of each.

Method 4 — Mobile (Android, iOS)

On mobile, the menu labels vary by Google app version, account, and region. This general path works on both platforms:

1. Open the Google app (not Safari or Chrome)
2. Profile icon in the top right → Settings
3. Search Labs → find the experiment related to AI Overviews → turn it off

If you don’t see the Search Labs experiment or the steps don’t work, a more reliable workaround:

• Set a web filter in your browser (see Method 2)
• Change your default search engine to Brave Search or DuckDuckGo (Safari → Settings → Search → Search Engine; Chrome on Android → Settings → Search engine)

Discover, Assistant, Gemini, and regular Search don’t have a single public “AI off” toggle — if you want to limit AI features broadly, go through the search engine, not a single setting.

Method 5 — For your own site: snippet directives selectively

If you see a drop in organic CTR on informational, non-brand queries in Google Search Console after AI Overviews rolled out, you can test limiting how selected blocks of content are used. Two levels, depending on the impact — don’t test site-wide; segment by query type.

Soft block — data-nosnippet on specific blocks

<div data-nosnippet>
  This paragraph won't be used in snippets or in AI Overviews.
</div>

data-nosnippet can stop Google from using a specific HTML block as a text snippet and, according to Google’s documentation, also limit how that block is used in AI Overviews. It isn’t a way to block a separate AI Overviews crawler — Google’s AI Overviews work with Googlebot and indexed content; the directive limits how the text is used, not whether it’s crawled.

Use it mainly on the opening answer block of long how-to guides, where the complete short answer usually sits.

Hard block — nosnippet site-wide

<head>
  <meta name="robots" content="nosnippet">
</head>

The standard form is <meta name="robots" content="nosnippet"> (or the Google-specific <meta name="googlebot" content="nosnippet">). It limits how your text content is used in snippets and AI Overviews across the whole page. It doesn’t remove the page from the results, and it doesn’t control every AI feature.

⚠️ Side effect: you’ll also lose classic featured snippets (position 0), which drive traffic for some queries. At a minimum, pilot it on a selected set of URLs before rolling it out site-wide and evaluate the impact in GSC — ideally test at the level of page groups and queries, not the whole site.

Maximum limit — max-snippet:0

<meta name="robots" content="max-snippet:0">

A hard limit of 0 characters per snippet. Setting max-snippet:0 sharply limits the ability to use the page’s text in snippets and can also limit how the content is used in AI Overviews. But it doesn’t remove the page from the index or guarantee zero presence across every AI format.

A decision framework for snippet directives

Never deploy snippet restrictions site-wide without segmenting:

• Test mainly on informational, non-brand queries where an AI Overview answers the query without a click and GSC shows a clear CTR drop
• Don’t apply them site-wide to brand, navigational, product, and conversion pages without evaluating each separately — on brand queries the goal is often to maximize control of the SERP, not block snippets; a quality brand snippet can be more valuable than a defensive block
• For YMYL content (health, finance, law), weigh the impact on brand trust — a missing snippet can send users elsewhere
• For e-commerce, the impact is often milder (the intent is transactional, not informational), but test anyway

Beware of AI Overviews “impact” claims with no methodology

In marketing discussions you’ll see claims like “AI Overviews cut site X’s traffic by YY%.” Before you draw conclusions for your own site, check:

• The data source (GSC, GA4, a Similarweb estimate, Ahrefs) — third-party tool estimates are approximations, not exact numbers
• Seasonality control (a year-over-year comparison, not month-over-month)
• Other changes (a core update, a redesign, technical changes, link building)
• Query-level data in GSC — a drop at the domain level can come from causes other than AI Overviews; segmenting by non-brand vs. brand and by intent (informational vs. transactional) is essential

Correlation between AI Overviews and a traffic drop isn’t automatically causation. A decision about snippet restrictions should come from GSC data at the level of queries and landing-page types, not from aggregated third-party estimates.

Which methods work and when

Recommendation by user profile:

• Regular user → try the Search Labs experiment + Brave/DuckDuckGo as a backup
• Power user who wants classic Google → the &udm=14 web filter as the default search in your browser (knowing Google can change the behavior)
• Privacy-first → Brave Search or DuckDuckGo permanently (check the opt-in AI features)
• Publisher with a proven CTR drop on informational, non-brand queries → data-nosnippet selectively after segmenting, with an A/B test
• Brand / e-commerce / YMYL → don’t apply snippet directives site-wide without a separate impact analysis of the brand SERP

What’s not recommended

Summary in five lines

1. There’s no guaranteed global off switch for Google AI Overviews — Google doesn’t offer one.
2. The Search Labs experiment — if you see it in your account, turning it off can limit AI features. But it isn’t a global switch.
3. The &udm=14 web filter is currently the most stable workaround for classic Google without AI Overviews. It isn’t an official directive — Google can change the behavior.
4. Brave Search or DuckDuckGo as your default search engine is the most reliable route to Google without AI Overviews.
5. For your own site, use snippet directives selectively after segmenting your queries, not site-wide. Before you implement, run an A/B test, segment brand vs. non-brand, and check the impact on featured snippets.

Publisher mini-method — how to decide on a snippet policy

If you’re dealing with AI Overviews as a site owner rather than a user, we recommend a five-step process instead of a site-wide rollout:

1. 1

   Pick a cluster of URLs

   Typically 10–30 informational, non-brand pages with a comparable query type (how-to guides, definition articles, comparisons). Don't mix them with brand and transactional URLs.

2. 2

   Filter queries in GSC

   Performance → Queries, filtered by brand vs. non-brand. A CTR drop on non-brand informational queries with steady or rising impressions is your signal.

3. 3

   Baseline the metrics

   Clicks, impressions, average position, and CTR for 4–8 weeks before the test — to capture seasonality.

4. 4

   Deploy on the segment only

   data-nosnippet on the opening answer block of 5–10 selected URLs, not site-wide. Watch for 30–60 days.

5. 5

   Evaluate after a few weeks

   Compare baseline vs. post-test, check the brand-query mix (a brand snippet is often an advantage, not a loss), and judge whether the CTR drop improved without unintended side effects.

Who this article is for

Where to get advice

This site is run by Sniper Design — if you’re dealing with AI Overviews as a business problem rather than a personal setting, we offer an AI visibility audit:

• For publishers and content sites — a query-level GSC audit, brand/non-brand segmentation, and a defensive snippet policy by content type
• For e-commerce — a check on whether AI Overviews is a real problem or just noise in aggregated data (transactional intent often doesn’t trigger the panel)
• For SEO/content teams — if you’re seeing a CTR drop after AI Overviews, we propose a test-first approach to snippet restrictions on selected URLs and measure the impact

We work on this with publishers, content sites, and lead-gen projects. For clients with informational content, we’ve typically found after segmentation that the CTR drop affected only a subset of their non-brand how-to queries, not the whole site — which fundamentally changes the priority of any defensive move.

Related

• AI SEO — the umbrella over SEO, GEO, and AEO — what AI Overviews is, how it works, and why it matters for SEO
• Decision matrix — when AI Overviews presence helps (citation share) vs. hurts (zero-click)
• AI SEO playbook — an implementation framework for publishers and content sites