You publish a well-researched article. A week later a partner site republishes it, a news aggregator picks it up, and a scraper copies it word for word. Then someone asks ChatGPT, Perplexity or Google's AI Mode a question your article answers, and the citation goes to one of the copies. Your work powered the answer, but someone else got the link.
This happens because AI search tools do not "know" which version came first. They retrieve pages from a search index, and when several pages carry the same text, the system has to pick one URL to show. Canonical signals, indexing choices, site authority and which copies a crawler can reach all influence that choice.
In this guide you will learn how duplicate and syndicated copies compete for AI citations, what Google currently recommends for syndication partners (it changed in 2023), how to handle scrapers, and a step-by-step process to protect attribution in 2026. If you run a news or media site, pair this with our wider guide to AI search for publishers.
Key Takeaways
- AI answers in Google draw on Google's index, so whichever URL Google treats as the canonical version is usually the one that can be cited.
- Google no longer recommends cross-domain canonical tags as the main fix for syndication; it says partners blocking indexing of your content is the most effective solution.
- Other AI search tools (ChatGPT search, Perplexity) run their own crawlers and indexes, so a copy that is open to their bots can still be cited even if Google picks your original.
- Publish first, make the original richer than the syndicated version, and require a visible link back to your URL in every syndication agreement.
- Audit citations regularly: search your key prompts and note which URL each AI tool credits.
Why duplicate copies compete for the same citation
Google's AI Overviews and AI Mode are built on top of Search. Google's own documentation says the same SEO fundamentals apply and that there is nothing special to add:
“There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.”
— Google Search Central, AI features and your website (Google Search Central)
That has an important consequence for syndication. When Google finds several near-identical pages, it clusters them and selects one canonical URL to represent the group. If a high-authority partner's copy becomes the canonical, the partner is the one eligible to appear as the source. Our guide to duplicate content in SEO covers how that clustering works for normal search; the same mechanics now decide AI citations too.
Outside Google the picture is more fragmented. OpenAI says OAI-SearchBot is the crawler used to surface websites in ChatGPT's search features, and Perplexity says PerplexityBot is used to surface and link websites in its results. Neither company publishes a canonicalisation policy for duplicates, so treat their behaviour as unknown and assume any accessible copy can be chosen. For a wider look at why AI tools prefer other domains, see why AI cites other sites instead of yours.
Three types of duplicate, three different fixes
| Type of copy | Who controls it | Main risk for AI citations | Best fix |
|---|---|---|---|
| Syndication partner (licensed republishing) | You and the partner, via contract | Partner's stronger domain becomes the canonical and gets cited | Partner adds noindex to the copy; visible credit link to your original |
| Your own duplicates (tracking URLs, print pages, HTTP/HTTPS, category paths) | You | Signals split across URLs; wrong version cited | 301 redirects, consistent internal links, self-referencing rel="canonical" |
| Scraped or copied content | Nobody you have an agreement with | Copycat URL occasionally chosen as canonical | Contact the host, file a DMCA request, strengthen your own signals |
| Guest posts and republished articles you wrote for others | The host site | Your own blog version competes with the host's | Decide which URL should win and point signals to it |
What Google now recommends for syndication
For years the standard advice was "ask the partner to add a cross-domain canonical tag pointing to you". Google changed its guidance in May 2023, as Search Engine Land reported at the time. Its canonicalisation troubleshooting page (last updated August 2026) now says the canonical link element is not recommended for avoiding duplication by syndication partners, because the pages are often very different. Instead:
“The most effective solution is for partners to block indexing of your content.”
— Google Search Central, Troubleshoot canonicalization issues (Google Search Central)
In practice that means the partner adds <meta name="robots" content="noindex"> (or the X-Robots-Tag HTTP header) to the republished copy. A noindexed page cannot be shown in Google Search, so it cannot be chosen as the canonical or cited in an AI Overview. A cross-domain canonical tag is still a valid hint, and it can help, but Google treats it as one signal among several. If you want a refresher on how canonical tags work, read our canonical tag explainer.
Note what noindex does not do: it does not stop OpenAI's or Perplexity's crawlers from fetching the page. If the partner wants to keep the copy out of ChatGPT search, it would also need to disallow OAI-SearchBot in robots.txt; the equivalent for Perplexity is PerplexityBot. Many partners will not agree to that, which is why the visible credit link matters so much (more below).
Step-by-step: protect attribution when you syndicate
- Publish on your own site first. Give crawlers time to discover and index your original before the partner goes live. Request indexing in Search Console for important pieces.
- Write the noindex requirement into the agreement. Ask for
noindexon the syndicated copy. If the partner refuses, ask for a cross-domainrel="canonical"to your URL as a fallback. - Require a visible credit line. "This article first appeared on [Your Brand]" with a followable link to the exact original URL. Semrush's syndication guide also recommends a credit and link back to the original.
- Syndicate a shorter or older version. Keep the full data tables, original charts and expert commentary on your page. AI systems that compare passages will find more unique, quotable material on the original.
- Keep your original self-canonical and well linked. Link to it from relevant hub pages so it is clearly the main version on your domain.
- Check the partner's implementation. View source on the syndicated copy, confirm the noindex or canonical is present, and check that the robots meta tag is not added only by JavaScript after load.
- Audit citations monthly. Ask the prompts the article targets in ChatGPT, Perplexity, Gemini and Google AI Mode and record which URL is cited. Our AI citation audit process explains how to log this consistently.
Dealing with scrapers and copycat sites
Google defines scraping in its spam policies as taking content from other sites, often through automated means, and hosting it to manipulate rankings. Usually the original wins, but Google admits it does not always:
“our algorithm may select a URL from an external site that is hosting your content without your permission.”
— Google Search Central, canonicalization troubleshooting, Troubleshoot canonicalization issues (Google Search Central)
Google's suggested response is to contact the site's host or file a DMCA request. Beyond that, the best defence is making your original unmistakably stronger: earlier publication, more internal links, more references from other sites, author information and unique elements a scraper is less likely to copy cleanly, such as embedded data visualisations. If you suspect hacking (unexpected cross-domain canonicals or redirects inserted on your site), treat it as a security incident first.
Signals that help AI tools identify the original
- A clear, consistent byline and publication date in both visible text and Article structured data.
- Original data, quotes and images that only exist on your URL.
- Mentions and links from other publications pointing to your URL, not the partner's. If you distribute announcements, the same logic applies to press releases in AI search.
- Fast, crawlable HTML so AI crawlers can actually fetch your page; blocked or slow originals lose to accessible copies.
How to check who is getting the citation
| Check | Where | What it tells you |
|---|---|---|
| URL Inspection, "Google-selected canonical" | Google Search Console | Whether Google treats your page or another URL as the canonical |
| Exact-phrase search of a unique sentence | Google and Bing | Which copies are indexed and which one appears first |
| Prompt testing with sources expanded | ChatGPT, Perplexity, Gemini, AI Mode | Which URL each tool credits for your content |
| Referral and AI-source traffic to the original | GA4 (see tracking AI traffic in GA4) | Whether citations translate into visits to your site |
| Server logs for OAI-SearchBot, PerplexityBot, Googlebot | Hosting / CDN logs (see finding AI bots in server logs) | Whether AI crawlers reach your original at all |
If Search Console shows a partner's URL as Google's selected canonical for your article, fix the agreement before publishing anything else with that partner.
Common mistakes
- Relying only on cross-domain canonicals. Google calls canonicals hints and no longer recommends them as the main syndication fix.
- Letting the partner publish first. Whoever is indexed first with stronger signals has an advantage you may never recover.
- Syndicating the full piece with every chart and table. You hand over all the unique, quotable material.
- Noindexing your own original by mistake. Google's canonical documentation specifically warns against using noindex to choose a canonical within a single site, because it removes the page from Search entirely.
- Assuming Google's behaviour applies to every AI tool. ChatGPT search and Perplexity use their own crawlers; check them separately.
- Never checking. Many publishers only notice the problem when traffic disappears.
Related Guides
- Paywalls and AI Search: Visibility Without Giving Content Away
- International AI Search: How Language and Country Shape AI Answers
- Tables, Lists and Specs: HTML Formats AI Extracts Accurately
Frequently Asked Questions
Should I stop syndicating content because of AI search?
Not necessarily. Syndication still builds reach and brand awareness. The goal is to make the partner copy non-indexable (noindex) and to keep the richest version of the content on your own URL.
Is a cross-domain canonical tag useless now?
No. It is still a valid hint and better than nothing. Google simply says that partners blocking indexing of the copy is the most effective solution, because syndicated pages often differ from the original.
Does noindex on a partner page stop ChatGPT from citing it?
Not reliably. Noindex is a search-engine indexing directive. OpenAI controls ChatGPT search crawling through the OAI-SearchBot user agent in robots.txt, so a partner would need to block that bot as well.
What if a scraper outranks me in AI Overviews?
Check the Google-selected canonical in Search Console, contact the host or file a DMCA request, and strengthen your original with internal links, mentions and unique material. Google says its algorithm may rarely pick a copycat URL.
Can I republish my own guest posts on my blog?
You can, but decide which URL should win. If the host should get credit, noindex your copy or point a canonical to the host; otherwise expect the two versions to compete.
Protect the credit your content earns
Syndication and AI search can work together, but only if the original is the version search engines and AI crawlers can find, index and trust. Put noindex and credit-link requirements in every agreement, keep your best material on your own URL, and audit citations every month. If you would like help auditing canonicals, partner copies and AI citations, our SEO team can review your setup. Request a quote to get started.
References
- Troubleshoot canonicalization issues (Google Search Central)
- How to specify a canonical URL (Google Search Central)
- Spam policies for Google web search (Google Search Central)
- AI features and your website (Google Search Central)
- Overview of OpenAI crawlers (OpenAI)
- Perplexity crawlers (Perplexity docs)
- Semrush: What Is Content Syndication? (7 January 2025)
- Search Engine Land: Google no longer recommends canonical tags for syndicated content (3 May 2023)
