Does AI-written content rank on Google?
Google has said since 2023 that it rewards good content however it is produced. The rule that actually bites is scaled content abuse, and the bigger risk now is not ranking — it is the click.
8 minute read
The short answer, and where it comes from
Google has not banned AI-written content and has said so plainly since February 2023, when Search Central published guidance stating that it focuses on rewarding high-quality content regardless of how it is produced. That position has not moved since. What Google says it cares about is whether a page is original, useful and made for people, and whether it demonstrates experience, expertise, authoritativeness and trustworthiness. That is the honest answer to the headline question, and it is why most of the advice circulating on this topic is wrong in both directions. AI content is not automatically demoted, and it is not automatically safe either. The thing that gets sites into trouble is a specific, named behaviour, and it pays to read the policy rather than the commentary about it.
What the guidance says, and what it conspicuously does not
The 2023 post draws its boundary in one sentence: using automation, including AI generation, primarily to manipulate search rankings is against Google's spam policies. Google's helpful-content documentation repeats the same idea from the other side, listing extensive automation used to produce content across many topics as a warning sign that you are writing for a search engine rather than a person. Now notice what is absent. There is no threshold for how much of a page a model may have written, no requirement to run a detector, no obligation to add an AI byline, and no demotion attached to the act of using a language model. Every enforceable line in Google's documentation is about intent and output, not authorship.
The rule that actually bites: scaled content abuse
The policy to read is scaled content abuse, on Google's spam policies page. It defines the offence as generating many pages for the primary purpose of manipulating search rankings rather than helping users, and the listed examples include using generative AI tools to produce many pages without adding value for users, scraping feeds or search results to spin out pages, stitching together content from different pages without adding value, and running multiple sites to hide how scaled the operation is. The framing is deliberate. This policy replaced an older rule about automatically generated content, and the rewrite moved the test from how something was made to what it does. A thousand thin AI pages and a thousand thin pages from cheap freelancers break exactly the same rule.
Does AI-written content actually rank? What has been measured
Originality.ai has run a continuous tracker since 2019, pushing the top 20 Google results for 500 informational keywords through its AI detector every two months, roughly 10,000 pages per sample. The share flagged as AI-generated rose from 2.27% in February 2019 to an all-time high of 19.56% in July 2025, and sat at 17.31% in mid-September 2025. The shape is more interesting than the level: the figure dipped to 7.43% after Google's March 2024 core update before climbing back. That pattern is consistent with Google enforcing against low-effort scaled output rather than against AI as a method. Treat the absolute numbers with care, though — detector scores are probabilistic and produce false positives on plain, formal human writing.
Why unedited drafts underperform even with no penalty involved
Set the policy aside and the median unedited AI draft still loses on the criteria Google documents. Its helpful-content guidance frames quality around three questions: who made this, how it was made, and why it exists. The who and the why are precisely where a raw draft has nothing to offer — no byline with real standing behind it, no first-hand experience, no reason to exist beyond occupying a keyword. Google also asks directly whether the use of automation is self-evident to visitors through disclosures, and whether you provide background on the role it played. None of that is a penalty. It is a fairly accurate description of what a model cannot supply on its own, and therefore of the part you still have to bring yourself.
The bigger risk in 2026 is not ranking. It is the click.
Ranking is only half the transaction, and the other half has moved. Pew Research Center tracked the browsing of 900 US adults through March 2025, covering 68,879 Google searches, of which 12,593 returned an AI summary. Where an AI summary appeared, people clicked a standard search result on 8% of page visits. Where none appeared, they clicked on 15% — nearly double. Clicks on the links inside the summary itself were rarer still, at 1% of visits. People were also more likely to end their browsing session entirely after a page with an AI summary, 26% of the time against 16%. Roughly 18% of searches in the sample produced a summary at all, and Wikipedia, YouTube and Reddit were the most cited sources.
Google reads the same trend very differently
Google disputes this picture, and the disagreement is worth understanding rather than picking a side in. On 6 August 2025 its head of Search, Liz Reid, published a post saying total organic click volume from Google Search to websites had been relatively stable year over year, that average click quality had risen, and that third-party reports of dramatic traffic declines rest on flawed methodologies, isolated examples, or changes that predate AI features. Reid defined a quality click as one where the user does not quickly bounce back. The post included no supporting data, which is the fair criticism of it. Both readings can hold at once: Pew measured what a panel did on individual searches, while Google is describing an aggregate only it can see.
What to actually do about it
The practical conclusion is unglamorous. Use AI for the parts a model is genuinely good at — drafting, restructuring, summarising, transcription — and put your own judgement, data and experience into the part that makes a page worth ranking. Publish fewer pages that each say something rather than many that repeat each other, because volume without value is the specific thing the spam policy names. If a reader might reasonably wonder how a piece was made, say so. And since AI summaries increasingly answer the shallow questions themselves, the queries worth chasing are the ones needing an opinion, an original number, or a demonstration. That argues for putting effort into video, where a clip of you showing the thing is not something a summary can paraphrase away. A tool like FrameOS makes turning one recording into that supply of clips a smaller job.
FAQ
Does Google penalise AI-generated content?
No. Google's published guidance since February 2023 says it rewards high-quality content regardless of how it is produced. What violates policy is scaled content abuse: generating many pages primarily to manipulate rankings rather than help users. Google's spam policies name generative AI as one way of doing that, alongside scraping and stitching, but the test is the output, not the tool.
Do I have to disclose that AI helped write an article?
There is no ranking requirement to. Google's helpful-content documentation asks whether the use of automation is self-evident to visitors through disclosures, and whether you give background on the role it played, suggesting disclosure is useful where someone might wonder how a piece was created. Treat it as an editorial decision about reader trust rather than a compliance step.
Can Google tell if content was written by AI?
Google has never published a detector or claimed to judge pages by an AI percentage; its stated test is quality and intent. Third-party detectors exist but are probabilistic — Originality.ai's own tracker works from a confidence threshold rather than a binary verdict, and formal human writing can trip them. Do not make publishing decisions from a detector score.
Are AI Overviews reducing traffic to websites?
The evidence conflicts. Pew Research Center found people clicked a search result on 8% of visits where an AI summary appeared, against 15% where none did, and were more likely to end the session afterwards. Google's Liz Reid said in August 2025 that total organic click volume has been relatively stable year over year, though that post published no figures.
Sources
- Google Search's guidance about AI-generated content · Google Search Central
- Spam Policies for Google Web Search · Google Search Central
- Creating helpful, reliable, people-first content · Google Search Central
- Amount of AI Content in Google Search Results (ongoing study) · Originality.ai
- Do people click on links in Google AI summaries? · Pew Research Center
- AI in Search is driving more queries and higher quality clicks · Google
Related pages
Keep reading
More AI →The best AI tools for social media, by the job you need done
Most AI tool lists are directories. This one is organised by job, because the useful question is not which tool is best overall, it is which step of your week you want to stop doing by hand.
9 minute readThe best AI transcription tools, judged on measured accuracy
On clean read speech almost every engine looks excellent. On meetings, accents and crosstalk the same models lose several points of accuracy — and the published numbers show exactly where.
8 minute readThe best AI writing tools for creators, and what each one is actually for
Most roundups list forty tools that all do the same thing. The useful split is between general assistants, the editing layer, and tools that write from material you already have.
8 minute readTurn one long video into a week of posts
300 credits for 3 days · no card.
Start free