Not for being AI-written. Google has said since 2023 that it rewards helpful content however it is produced, and its spam rules target pages mass-produced to game rankings, by machines or people. AI drafts are fine; unchecked, generic volume is the risk.
No. Google does not penalise content for being written with AI. Its position since February 2023 has been that it rewards high-quality content however it is produced, and that appropriate use of AI or automation is not against its guidelines. What it acts against is scaled content abuse: many pages generated mainly to manipulate search rankings, whether a model, a team of writers or both produced them.
So the useful question for a business owner is practical: which uses of AI produce pages that help customers, and which produce the pages Google's systems are built to filter out? Here is what Google has actually said, what changed in 2024 and 2025, and a workflow that keeps you on the right side of it.
- less low-quality, unoriginal content in Google results after its March 2024 changes, by Google's count
- 45%
- less low-quality, unoriginal content in Google results after its March 2024 changes, by Google's count
- Google core updates in 2025: March, June and December
- 3
- Google core updates in 2025: March, June and December
- the year Google said appropriate use of AI is not against its guidelines
- 2023
- the year Google said appropriate use of AI is not against its guidelines
01
What Google has actually said
Google's guidance has been consistent for three years. What changed is how firmly it enforces the spam side.
Dec 2022
Experience joins E-A-T
Google's rater guidelines add first-hand experience, such as actually using a product, to expertise, authoritativeness and trust.
Feb 2023
Quality over method
Google says it rewards high-quality content however it is produced, and that using automation mainly to manipulate rankings is spam.
Mar 2024
Core update and new spam policies
Helpfulness moves into Google's core ranking systems, and scaled content abuse becomes a named spam policy.
Apr 2024
45% less low-quality content
Google reports the March changes cut low-quality, unoriginal content in results by 45%, against the 40% it had expected.
May 2025
Guidance on generative AI content
Google publishes a page on using generative AI on websites: useful for research and structure, risky for mass-produced pages.
2025
Three core updates
In March, June and December, plus a spam update from late August.
Appropriate use of AI or automation is not against our guidelines.
02
What actually gets sites into trouble
Google defines scaled content abuse as generating many pages for the primary purpose of manipulating search rankings rather than helping users. Its examples include using generative AI tools to produce many pages that add no value, scraping feeds or search results, stitching together content from other pages, and pages that make little sense to a reader but contain search keywords.
Enforcement works in two ways. Google's core ranking systems, which it updates several times a year, are tuned to show more helpful results. Separately, when a human reviewer finds pages that break its spam policies, Google can apply a manual action and report it in Search Console's Manual actions report and message centre. The categories include thin content with little or no added value, and major spam problems.
03
What Google rewards instead
Google's helpful content guidance asks whether a page offers original information or analysis, goes beyond the obvious, would be more useful than other pages in the results, and was written or reviewed by someone who clearly knows the subject. It also says Google has no preferred word count. Behind those questions sits E-E-A-T: experience, expertise, authoritativeness and trust. Google says E-E-A-T is not itself a specific ranking factor, and that of the four, trust matters most.
Experience is the part a language model cannot supply. A model can summarise what every site already says about knee pain or GST registration. It cannot tell a reader what a Kochi physiotherapy clinic sees every monsoon, or which mistakes a Thrissur accountant's clients make on their first filing. Those specifics are what make a page worth more than the ten others on the same topic.
Google also suggests telling readers how content was made where they might reasonably wonder: accurate author bylines, and where it helps, a note on how AI was used. The guidelines for Google's human search quality raters, last revised in September 2025, have sections on scaled content abuse and on main content created with little to no effort, originality or added value. Raters do not set rankings; Google uses their ratings to measure how well its systems work.
04
Where AI helps, and where it hurts
In practice the line falls between AI as a tool for people who know the subject and AI as a replacement for knowing it.
| Use of AI | Where it stands | Why |
|---|---|---|
| Researching a topic and outlining a page | Fine | Google calls generative AI particularly useful for research and for adding structure to original content |
| Drafting a page that someone then checks and fills with specifics | Fine, if the result helps readers | Google rewards quality however content is produced |
| Dozens of near-identical pages, such as one per town | High risk | Generating many pages that add no value is a named example of scaled content abuse |
| Rewriting or combining other sites' articles | High risk | Stitching together content from other pages without adding value is a named example |
| AI-made product images shown through Merchant Center | Allowed, with labels | Merchant Center requires metadata marking them as AI-generated |
05
A workflow that scales safely
This is the process we recommend to clients who want to publish regularly without a large team.
- Start from real questions. Pull them from WhatsApp chats, calls and enquiry forms, not from a keyword tool alone.
- Let AI research and outline, then check every claim and source it offers, because drafts can contain confident errors.
- Draft, then add what only you know: prices, local details, photos of your own work, the questions customers really ask and the mistakes you keep seeing.
- Verify and cut. Check every number, date and claim, and delete filler and repeated points.
- Sign it. Publish under a real person's name with a short bio, and say how AI was used where a reader might wonder.
- Review the extras. Titles, meta descriptions and image alt text appear in search results too.
- Prune. Update or remove pages that no longer help anyone.
What to do this month
- Open Search Console's Manual actions report and confirm it shows no issues.
- List everything published in the last year and flag pages no customer would miss.
- Add one thing only your business knows to each of your five most important pages.
- Put real author names and short bios on articles.
- Write a one-paragraph AI policy for your team: what AI may draft, who checks it, and what never goes live unchecked.
For how this connects to AI answers, see why SEO is still the qualifying round. If you would like a second opinion on your content, ask for a free audit.