Do AI-Generated Blog Posts Rank on Google? What the Data Says

6 min read·30 September 2026

Yes, AI-generated blog posts can rank on Google, and some reach the top three results. Google doesn't rank or demote a page because AI wrote it. It judges whether the page is useful. Still, the 2026 data is clear that pages written or heavily shaped by people take most of the top spots, and posts that are mostly AI tend to get fewer impressions.

So the real question isn't whether you're allowed to use AI. It's whether an AI-generated blog post can be good enough to beat the pages already ranking for your searches. Below we go through Google's position, three recent studies and what they mean for a small business blog.

What Google's AI content policy actually says

Google's current guidance is short. Its page on generative AI content says AI "can be particularly useful when researching a topic, and to add structure to original content." The warning comes next. Using AI to churn out lots of pages "without adding value for users" may break its spam policy on scaled content abuse.

"Scaled content abuse" means publishing large numbers of pages mainly to catch search traffic, not to help anyone. The rule doesn't care whether a person, a script or an AI wrote them. What matters is the volume and how little each page adds.

The same guidance asks you to focus on "accuracy, quality, and relevance," and that includes the parts people forget: title tags, meta descriptions, structured data and image alt text. It also suggests telling readers how a piece was made when that would help them.

Google's earlier guidance, quoted by Ahrefs, says the problem is content made "with the primary purpose of manipulating ranking in search results," not AI as such. We cover the spam rules in more detail in what Google's guidelines say about penalising AI content. This post is about the evidence.

What the 2026 ranking data shows

Three large studies came out this year. They measured different things, so it's worth looking at each one on its own terms.

Ahrefs: fully AI pages reach the top three, but rarely

In July 2026, Ahrefs published a study of pages drawn from 100,000 search results. It ran them through its own AI detector. Among pages in positions 1–3, 5.3% came back as 100% AI-generated and 9% as at least 80% AI.

So fully AI pages do reach the top. They're a small minority, though. Pages with less than 50% AI content made up 82.2% of top-three rankings, and 54.7% had less than 20%.

Ahrefs also looked at indexing, which means whether Google stores a page and can show it in results at all. Pages with little AI content had an indexation rate of 49.28%, against 40.35% for pages that were mostly AI. That's a gap, but it's no ban.

The biggest difference was in impressions, meaning how often a page appeared in search results. Pages with low or moderate AI content received "2–3x the impressions" of mostly-AI pages. Ahrefs found no sudden collapse in impressions for AI pages over time, either.

Semrush: human-written pages hold position one

Semrush looked at 42,000 blog posts ranking for 20,000 keywords and published the results in April 2026. Posts classed as purely AI-generated held the top spot "just 9% of the time," while posts classed as human-written held it 80% of the time.

The detail matters more than the headline figure. According to Semrush, "from position 5 onward, the gap between human-written and AI-generated content is relatively narrow." AI content holds its own on page one and falls behind at the very top.

The same study surveyed SEO teams. 70% said speed was the main benefit of AI, but only 19% said it made their content better. That's a fair summary of where most AI blogging sits today.

Graphite: half of new articles are AI, far fewer get seen

Graphite's research found that primarily AI-generated articles made up 49.9% of new articles published in Q1 2026, roughly level with human-written ones. It also says a separate study of theirs found these articles "largely do not appear in Google and ChatGPT."

Put the three studies side by side and you get a lot of AI-written content, some of it ranking, and most of the top spots still going to pages with clear human input.

Is AI content bad for SEO, or is it just bad content?

Mostly the second. Ahrefs puts it plainly. Google isn't against AI content, "it is against bad content," and the confusion comes from how often the two overlap.

Ahrefs lists where basic AI content usually falls short. It repeats what's already out there, leaves out links, images and first-hand experience, reads like an essay and gets facts wrong. Any of those would hurt a human-written post too. They just turn up far more often when a model writes on its own.

Semrush makes a similar point about its detector: "It reads the finished product." A post that's generic and has no sources looks the same whether it came from a tool or from a tired writer on a Friday afternoon.

There's one more pattern worth knowing about. Ahrefs mentions "Mount AI" traffic graphs, where a site publishes AI content at volume, climbs quickly and then drops a few months later. That's what scaled content looks like when it goes wrong. Publishing two careful posts a month isn't the same thing.

How to run an AI-generated blog that can rank

Say you run a joinery business in Galway and you use AI to draft a post on "oak vs ash worktops" (this is a hypothetical example). A draft that only restates what the top ten results already say gives Google no reason to rank you above them. Here's what closes that gap:

1. Start from a real search. Write for queries people already find you for. Our guide to finding blog topic ideas in Search Console shows how.
2. Add what only you know. Include your prices, your lead times and the mistake customers make every time. The studies suggest this is where human-led pages get their edge.
3. Check every claim against its source. Wrong facts are one of the most common failings Ahrefs lists. If a number has no source, cut it.
4. Fix the small things too. Google's guidance names title tags, meta descriptions, structured data and alt text. Those are easy to rush when a tool writes them.
5. Publish at a pace you can review. Read every post before it goes live. If you can't keep up with the volume, you're drifting towards scaled content.
6. Measure, then improve. A new post often starts on page two. Our post on striking distance keywords, meaning searches where you already rank in positions 4–20, explains how to push those pages up.

This is how we built Blog AI. It starts from your Search Console data, links every claim to its source and leaves the final read to you. It can't give you the customer stories and pricing that make a post worth ranking. You still have to add those.

What these studies can't tell you

Every study here uses an AI detector, and detectors guess. Ahrefs says outright that its "AI content detection is not perfect" and won't work the same way Google does, if Google checks at all. Graphite used three detectors to reduce errors, and even then the figures are estimates.

They also show correlation, not cause. Ahrefs points out that sites relying heavily on AI may simply be newer or have less authority to begin with. A new site is at a disadvantage whoever writes its posts.

And none of them promise anything about your own site. Nobody can honestly promise you rankings or traffic, and that includes us.

AI can write a post that ranks, but the posts that win the top spots are the ones where a person added knowledge, checked the facts and cut the filler. Your next step: pick one query you already show up for in Search Console, and write or update one post for it that includes at least one thing only your business knows.

Researched and drafted with Blog AI, checked against its sources, and edited before publishing. How the checks work