Google's UGC spam guidance: what AI tools should index

By Redacción Usanzia · Published 27 October 2026

Since 26 October 2026, Google has asked tools that publish shared pages, generated content or web apps to index only what's meant for Search. For shared chats, artifacts and user-built apps, it recommends noindex or a crawl block by default; if those pages are already indexed, use noindex without blocking them in robots.txt, or Google won't see the rule.

See the steps

What has Google changed in its user-generated spam guidance?

Google has expanded its guide Prevent user-generated spam on your site and platform to explicitly cover hosting platforms and tools that let users publish shared pages, generated content or web apps. According to the Search Central documentation changelog, the update went live on 26 October 2026 and adds guidance on blocking the indexing of link-shared content and on monitoring abuse in Search Console.

The guide notes that these platforms can be as open to abuse as an unprotected comment form: spammers may create large numbers of pages or sites that break Google's spam policies and add little or no value to the web. There are three main changes:

• Wider scope: the introduction now names hosting platforms and tools that let users publish shared pages, generated content or web apps.
• A new section, Only index content meant for Search.
• Search Console monitoring: verify the site or subdomain to get alerts, and keep an eye on the indexing and performance reports.

Google's stated reason is to help platforms keep user-created pages that aren't meant for Search out of the index, and to detect spam earlier.

Google warns tools on user-generated spam guidance and adds a new section for only index content meant for Search

Post by @rustybrick · View on X

What does the new "Only index content meant for Search" section say?

It asks services that let users create shareable pages, generated outputs or interactive tools to decide whether that content is meant to be found through search engines or only reached by people who have the link. Where pages exist for direct sharing, collaboration or utility rather than public discovery, Google recommends applying the noindex rule (via the robots meta tag or the X-Robots-Tag HTTP header) or blocking crawling in robots.txt by default.

"By default" is the important part: the guidance isn't about cleaning up later, it's about making sure anything not meant for Search never enters the index in the first place.

Does this cover shared chats, artifacts and AI-built apps?

In our reading, yes. A shared chat, an artifact (a piece of content or mini app produced by an AI assistant that can be published via a link) or a web app built with an AI tool fits the categories the guide describes: shareable pages, generated outputs and interactive tools. Google, however, doesn't name any specific tool.

Others in the industry read it the same way. Lily Ray wrote on X that her first thought was "Claude artifacts, ChatGPT pages/sites" and other social and AI tools that make it easy to churn out indexable UGC at scale (her post, 26 October 2026). Marie Haynes asked whether new manual actions were coming for AI tools and app-building platforms, and argued that Google is putting "the burden of curation" onto tool makers (her post, 26 October 2026).

Two points we checked against the documentation:

• No new manual actions are announced. The guide tells you to verify your site in Search Console so you get an email if Google issues a manual action or detects security issues. It doesn't say new or tool-specific actions are on the way.
• The spam policies page didn't change. The 26 October changelog entry only mentions the prevention guide, and the spam policies page shows 28 August 2026 as its last update. The spam policies page already had a user-generated spam section linking to it.

What do Google's spam policies say about this kind of content?

Google's spam policies define user-generated spam as spammy content that users add to a site through a channel intended for user content, often without the owner knowing. Examples include spammy accounts on hosting services anyone can sign up for and spammy files uploaded to file hosting platforms.

The same page defines scaled content abuse as generating many pages mainly to manipulate rankings rather than help users, "no matter how it's created". One of its examples is using generative AI or similar tools to produce many pages without adding value for users. If you host that kind of content, Google says to exclude it from Search.

For an AI product, the practical takeaway is that the risk isn't just what the company publishes, but what its users can turn into public, indexable URLs.

noindex or robots.txt: which should you use for shared pages?

For pages that shouldn't appear in Google, noindex (as a meta tag or an X-Robots-Tag header) is the safer choice, because a robots.txt block doesn't guarantee the URL leaves the results. The prevention guide accepts both, but Google's documentation on robots.txt and noindex draws some important distinctions:

A URL blocked in robots.txt but linked from other pages can still show in results without a description, so for chats, artifacts and apps shared by link, noindex is the way to go.

The mistake to avoid is doing both on the same URL. Google warns that if a page is blocked by robots.txt, the crawler never sees the noindex rule and the page can still appear in results. So if shared pages are already indexed, our advice is to add noindex and not block them in robots.txt until they've dropped out. Google also notes that it doesn't support noindex inside robots.txt. And the rule should live in the HTML your server returns or in the X-Robots-Tag header. We recommend not adding or removing it with JavaScript, because according to Google's JavaScript SEO basics, when Google finds noindex it may skip rendering and JavaScript execution.

Which generated content deserves to be indexed?

Only content a user has deliberately published to be found, and that stands on its own; everything else should stay out of the index by default. Google's guide offers two levers:

• User reputation. Google suggests adding noindex to posts from new users with no reputation on the platform and allowing indexing once they've earned it, which it says strongly discourages spammers.
• Links in untrusted content. Google recommends adding rel="nofollow" or rel="ugc" to all links in untrusted content. Its guide to qualifying outbound links adds that you can drop the attribute for users who have consistently made high-quality contributions.

Our rule of thumb for AI tools: indexable only if the user explicitly publishes it as a public page (generating a share link isn't enough), the page makes sense without the chat context, and it has passed some form of review, automated or human.

How do you monitor abuse in Search Console?

Verify the site or subdomain where user pages live and check the indexing and performance reports regularly. Google's guide spells out what to look for:

• unexpected spikes in indexed pages, impressions or queries that don't match your platform's purpose (Page Indexing and Search performance reports);
• your own creation and indexing metrics, with alerts for anomalies such as spikes in new public pages or a small number of users generating a large share of indexable URLs;
• spam signals such as redirects, lots of ad blocks, certain spammy keywords or large chunks of encoded JavaScript (the site: operator or Google Alerts can help);
• sudden traffic spikes in server logs, plus phishing or malware pages (Google mentions the Safe Browsing API).

The guide also keeps its long-standing preventive measures: a clear abuse policy shown at sign-up, a way for trusted users to report content, manual moderation for suspicious interactions, blocklists, and reCAPTCHA or similar checks against automated account creation.

Trust and evidence

• Primary sources: Google's guide Prevent user-generated spam on your site and platform, the 26 October 2026 entry in the Search Central changelog, Google's spam policies, the noindex documentation, the introduction to robots.txt and the guide to qualifying outbound links. For the JavaScript and removals points, Google's JavaScript SEO basics and the Removals tool help page. All checked on 27 October 2026.
• Original testing: we haven't run our own test for this article (we don't operate a platform that publishes user pages).
• Method and limitations: we read Google's English documentation and cross-checked it against Search Engine Roundtable's coverage. We didn't compare the previous version of the guide ourselves: what "changed" relies on Google's changelog and that coverage. That the spam policies didn't change on 26 October rests on that changelog entry and on the last-updated date shown on the policies page (28 August 2026). Applying the guidance to shared chats, artifacts and AI-built apps is our reading, not Google's claim; Google names no tools. Lily Ray's and Marie Haynes's views are their own and are linked to their posts.