Resources

Can you publish AI-written content? What Google says

Written by
Outseen
Updated

Short answer

Yes, you can publish content made with AI. Google assesses the quality of content, not how it was produced. Mass-producing pages with no added value does fall under the spam policy on scaled content abuse, however those pages were made.

Yes, you can publish content made with AI. Google assesses the quality of content, not the way it was produced. The line is drawn at mass-producing pages with no added value: that falls under Google’s spam policy on scaled content abuse, whether those pages were made by AI, by people or by a combination of the two.

What exactly does Google say?

In February 2023, Google published guidance on AI-generated content. The core of it: Google looks at the quality of content, not at how it came about. Appropriate use of AI or automation does not breach its guidelines. Using automation with the main aim of manipulating search rankings does.

Google draws a comparison there with the rise of mass-produced content written by people, roughly ten years earlier. Then, too, Google didn’t ban human-written content; it improved its systems to reward quality.

According to Google, AI gives you neither an advantage nor a disadvantage. Content that is helpful, original and reliable can perform well. Content that isn’t probably won’t, regardless of who or what wrote it.

Where is the line?

In March 2024, Google introduced a new spam policy on scaled content abuse. By this, Google means creating large numbers of pages with the main aim of manipulating rankings rather than helping users. The policy applies regardless of how the content is produced.

In its spam policies, Google gives examples including:

  • using generative AI to create many pages with no value for users;
  • scraping content from other sites or search results and automatically rewriting, translating or swapping in synonyms;
  • stitching together content from different pages without adding anything;
  • creating pages that mean little to a reader but do contain search terms.

Since May 2026, the Search Central changelog also states explicitly that the spam policies apply to generative AI responses in Google Search.

What counts is the purpose and the result. A business owner who uses AI to produce a first draft, then adds their own knowledge and checks it, is fine. Someone who generates a hundred location pages that differ only in the town name is not.

Where is the line? Three situations

Within the guidelinesScaled content abuse
Pages per locationOne page per branch, with its own address, team and rangeMany pages where only the town name changes
Product descriptionsAn AI draft, filled out with sizes, materials and your own experience of the productText from other sites rewritten automatically
TranslationsA translation checked by someone who knows the languageAnother site’s content translated automatically and published
Based on the examples in Google’s spam policies. The purpose and the result set the line, not the tool.

The difference lies in what the reader gets out of it. An AI draft only becomes useful once something is added that only you know:

An AI draft filled out with your own knowledge

VoorOur e-bikes are of high quality and suitable for everyone. We offer excellent service and a wide choice.

NaAt The Spoke Cycle Shop we adjust every e-bike to your height before you take it home. If you are torn between two models, you can test-ride both on the same day.

The first text fits any bike shop. The second says what this shop does. The Spoke Cycle Shop is made up.

What do you check before publishing?

In its guidance on generative AI content, Google explains that language models don’t look up facts; they predict a likely sequence of words. That is why AI output can contain errors, also known as hallucinations. Google asks you to check all AI content by hand before publishing.

According to Google, that check also covers parts you are less likely to read:

ElementWhy it matters
Page title (<title>)Can appear in search results
Meta descriptionCan be shown as the description under your link
Structured dataMust match the visible content and Google’s rules
Image alt textCan be used in search results

Above all, check the things AI doesn’t know about you: your prices, terms, delivery times, warranties, and what you do and don’t supply.

From AI draft to published page

  1. BriefWrite down who the page is for, which question it answers and which sources are reliable.
  2. DraftLet AI produce a structure or first version based on that brief.
  3. Fill outAdd what only you know: experience, examples, how you work, your own photos.
  4. FactsCheck every claim against its source, and every figure, price and condition against your own records.
  5. MetadataReview the title, meta description, structured data and alt text.
  6. PublishPut the page live under the name of someone who stands behind the content.
The steps follow Google’s guidance on generative AI content. Who does which step is up to you.

Do you need to say you used AI?

That depends on what a reader expects. Google recommends explaining how content was made where someone would reasonably wonder. That is more likely with a test or a comparison than with a page of opening hours.

Google advises against two things. Crediting AI as the author is, according to Google, probably not the best way to make clear that AI helped write a piece. And Google calls fabricated authors, such as AI portraits attached to invented names, deceptive. Name a real person as the author, someone who stands behind the content. More on this is in E-E-A-T: showing experience and expertise.

Extra rules for online shops

If you sell through Google Merchant Center, additional rules apply. According to Google, AI-generated images must contain metadata that marks them as AI imagery (the IPTC value TrainedAlgorithmicMedia). AI-generated titles and descriptions in your product data must be specified separately and labelled as AI-generated.

Do and don’t at a glance

DoDon’t
Use AI for research, outlines and structureLet AI invent facts you don’t check
Have a real author check and sign off the contentCredit AI or an invented person as the author
Add your own knowledge, experience and examplesHave other people’s content rewritten
One good page per topicMany near-identical pages for every search variation
Review titles, descriptions and schemaPublish metadata unseen

How Outseen approaches this

Outseen uses AI as a working aid. A writer creates content from your briefing and sources, a reviewer checks it and you give feedback before we publish. Our specialists remain responsible for the work, and we never treat AI output as measurement evidence. How to write content AI assistants can use is covered in writing content AI can cite. What else we can do for your customer questions is on our AEO page.

What does this mean for your business?

You do not have to avoid AI, and you do not have to hide it. Do decide who checks a text before it goes live, and what that person reviews: facts, prices and terms, and the metadata you cannot see on the page itself. On a small website that can be one person; what matters is that it happens for every page, including a quick edit.

Look critically at the number of pages you plan to create, too. A plan for dozens of near-identical pages is a signal to stop and ask what each page adds for a reader. One page that answers a customer question well helps visitors, Google and AI assistants more than ten variants that do not. How Google assesses findability in general is explained in What is SEO and how does it work?

Sources

Read next

FAQs

Can Google tell that a text was written with AI?

We don’t know, and for the rules it makes little difference. Google says it assesses the quality of content, not how it was produced. A helpful text made with AI can therefore perform well; a poor text written by a person can’t. Google does have systems that detect spam, however it was made. So the main question to ask yourself is whether your content helps a reader.

Can I use AI to write product descriptions for my online shop?

Yes, as long as you check them and they are correct. Google asks you to fact-check all AI output by hand before publishing. Google Merchant Center has extra rules: AI-generated titles and descriptions in your product data must be specified separately and labelled as AI-generated, and AI images need specific metadata. So don’t let AI invent sizes, materials or prices; those come from your own product data.

Should I name ChatGPT as the author if I used it?

No, Google advises against it. Google writes that crediting AI as the author is probably not the best way to let readers know AI was involved in creating the content. Name a real person as the author, someone who is responsible for the content. Explain on the page or in your editorial policy how you use AI, where a reader might wonder how a piece was made.