

AI content SEO in a two-week spam update: sift the pile before Google does
At 09:15 Pacific time on 24 September, Google started a spam update it says may take up to two weeks. This year's three earlier spam updates were done in under three days each.
By Katie Delaney / 2026-09-28 / 12 min read

What Google rolled out, and why the length matters
| Item | Value |
|---|---|
| Low AI content | 49.3 |
| Very high AI content | 40.4 |
A fox does not panic at the first rustle in the brush. It listens for how long the rustling lasts. On 24 September, the Google Search Status Dashboard logged the September 2026 spam update, applying globally and in all languages, with a rollout that may take up to two weeks. On Monday morning it was still running.
The length is the news. As Search Engine Journal's SEO Pulse noted, this is the fourth spam update of 2026, after March, June and August, and Google gave each of those a window of a few days. Google's John Mueller told Search Engine Roundtable it is likely to take longer than some of the previous ones.
What Google has not said is just as important. Search Engine Journal reported that Google has not said whether the update targets a specific spam policy or tactic. Plenty of commentary has linked it to scaled AI publishing. That link is commentary, not Google's statement, and any AI content SEO plan built on a rumour is a plan built on sand.
What spam updates are for
Google's own explainer on spam updates describes them as improvements to always-on detection, naming SpamBrain as its AI-based spam-prevention system. It also sets expectations for recovery: sites that fix problems may improve as automated systems learn over a period of months that the site complies. Months, not days. Plan your AI content SEO accordingly, with the patience of a fox on a long prowl.
Is AI content bad for SEO? What the policies actually say
The Top Search Updates & News of the Week are Here [From #SEOFOMO - Sep 27, 2026]. Google September 2026 Spam Update Has Been Unleashed. Google announces web multimodal Search performance reporting in Search Console.
Aleyda SolÃs put the update at the top of her weekly roundup on Sunday, beside Search Console's new multimodal reporting. Practitioners are watching it closely, and the first question clients ask is always the same: is AI content bad for SEO now?
The honest answer starts with Google's spam policies, last updated on 28 August 2026. They define scaled content abuse as generating many pages primarily to manipulate rankings rather than to help users, typically unoriginal content with little or no value, no matter how it is created. Generative AI is the first example listed, next to scraping, stitching and spinning.
The same page now defines spam to include attempting to manipulate generative AI responses in Google Search, a line every generative engine optimisation vendor should read twice. It also covers expired domain abuse and site reputation abuse, where third-party content rides a host's ranking signals. The target is the purpose and the pattern. The tool is incidental.
Does Google penalise AI content by itself?
People type "does google penalize ai content" into Google itself, and Google has answered it in writing. Its 2023 post on Google Search and AI-generated content says appropriate use of AI or automation is not against its guidelines, while using automation primarily to manipulate rankings is. Its current guidance on generative AI content repeats the point: generating many pages without adding value for users may violate the scaled content abuse policy.
Five hundred city pages
One template, a model filling in the town name and three generic paragraphs, published in a weekend to catch local queries.
Forty pages that earn their place
A model drafts structure from your own data; an expert adds local evidence, prices and photos; each page answers a question nobody else answers.
That is the whole of AI content SEO in one comparison. The model is the same. The first version is scaled content in the sense Google's policy describes. The second is research and drafting help, which Google explicitly allows. The difference is whether a person with knowledge added something a searcher could not get anywhere else.

What the data says about ai generated content SEO
| Item | Value |
|---|---|
| Mar 2026 spam | 0.8 days |
| Jun 2026 spam | 2 days |
| Aug 2026 spam | 2.7 days |
| Sep 2026 spam window | 14 days |
| May 2026 core | 11.9 days |
| Aug 2025 spam | 26.6 days |
Those durations come from the dashboard's ranking release history. The closest precedent for a long spam rollout is the August 2025 spam update, which ran for 26 days. A long window means more time for rankings to wobble, and more time for teams to misread normal volatility as a penalty.
On AI content itself, the best public data is correlational. Ahrefs' July 2026 study sampled a million pages from the top ten of 100,000 results pages, ran about 150,000 through its AI detector, and found AI use correlated with lower impressions, lower positions and lower indexation. It also found 5.3% of top-ranking pages scored as fully AI-written. There is no cliff, only a slope.
For ai generated content SEO, that slope is the useful finding. Heavily automated pages tend to be thinner, and thin pages tend to do worse, with or without a spam update. The ranking system does not need to detect a model to reward the page that says something new.

Measure carefully while two things move at once
Measurement is harder this month than usual. Google's Search Console help on how clicks and positions are counted says an AI Overview occupies a single position, all its links share that position, and a click to an external page in the Overview counts as a click. Search Console also added multimodal search data on 24 September, the same day the spam update began, covering Lens, Circle to Search and image uploads.
AI Overviews are moving traffic too. A preregistered field experiment posted to arXiv in August, not yet peer reviewed, found assigning 1,100 participants to an AI Mode-only experience cut click-through to external sites by 18.8 percentage points. Separate the spam update from the AI Overview trend before you blame either.
Five checks for AI content SEO while the update rolls
The practical move this fortnight is not to delete pages in a panic. Annotate 24 September in every report, keep daily Search Console exports, and hold structural changes until the dashboard marks the rollout complete. Meanwhile, run five quiet checks across the den.
List every page type produced from one template at volume, such as location, comparison or glossary pages.
For each type, write down what a searcher gets there that they cannot get on a competitor's page.
Review partner, coupon or sponsored sections against the site reputation policy.
Google says accuracy extends to titles, descriptions, structured data and alt text generated by automation.
Use branded and non-branded filters to separate spam-update movement from AI Overview effects.
The metadata check is the one most teams skip. Google's generative AI guidance says the focus on accuracy includes title elements, meta descriptions, structured data and image alt text when they are generated automatically. A thousand machine-written titles that all say the same thing is a pattern a spam system can see from a long way off.
Where AI genuinely helps without risk
- Research and outlines
- Structuring your own data
- Drafts an expert rewrites
- Translation with review
- Metadata you then check
Every one of those uses keeps a knowledgeable person between the model and the page, which is the heart of safe AI content SEO. That is the line Google's documents keep drawing, and it is the line that protects a site through this update and the next one.

After the update: build AI content SEO that compounds
When the dashboard marks the rollout complete, compare the fortnight after with the fortnight before, by page type rather than by site. If a template family fell and your best original pages held, you have your answer and your roadmap. If everything moved together, look first at AI Overviews and seasonality before you look at spam.
Recovery, if you need it, is slow by design. Google says improvement can come as its systems learn over months that a site complies, and that ranking benefit from spammy links, once neutralised, is lost rather than restored. That is an argument for fixing the pattern now rather than waiting to see whether the update found you.
The durable version of AI content SEO looks unglamorous. Fewer pages, each with a reason to exist. Original numbers, named sources and dates in the prose. Models used for the grunt work, experts used for the judgement. It is slower to produce and far faster to trust, which is exactly the trade Google keeps signalling it wants.
Google is not hunting the tool. It is hunting the pattern.
If you would rather have a guide for your AI content SEO through the thicket, our SEO and GEO team runs template audits and update reads like this one, and our content marketing team builds the expert-led pages that replace thin ones. Read our earlier guide on checking whether an update hit you, and our note on AI Overview tracking, before you draw conclusions. And if you want to talk it through, get in touch.
Frequently asked questions
What is the September 2026 Google spam update?
It is a Google Search spam update released on 24 September 2026 that applies globally and in all languages. Google said the rollout may take up to two weeks, much longer than the few days given to its three earlier 2026 spam updates. Google has not said which spam policy or tactic it targets.
Does Google penalise AI content?
Not for being AI. Google says appropriate use of AI or automation is not against its guidelines. What its spam policies target is scaled content abuse: many pages generated mainly to manipulate rankings without adding value, however they were produced. Quality and purpose matter, not the tool.
Is AI content bad for SEO?
It can be if it produces thin, unoriginal pages at volume. Ahrefs' July 2026 study found higher AI content levels correlated with lower impressions, positions and indexation, but found no hard cutoff. AI used for research and drafts, with an expert adding original value, is allowed and can perform well.
What counts as scaled content abuse?
Google defines scaled content abuse as generating many pages primarily to manipulate search rankings rather than help users, typically unoriginal content with little value, no matter how it is created. Examples include mass-generated AI pages, scraped or spun feeds and stitched content from other sites.
How long does it take to recover from a spam update?
Google says sites that fix policy problems may improve as its automated systems learn over a period of months that the site complies. Ranking benefit from spammy links, once neutralised, is lost rather than restored. Fix patterns early and expect slow recovery rather than a quick rebound.
How do I tell a spam update drop from an AI Overview drop?
Annotate the rollout date, compare page types before and after, and use Search Console's branded and non-branded filters. If template pages fell while original pages held, suspect the spam update. If informational queries lost clicks while impressions held, AI Overviews are the likelier cause.
The Google algorithm update lands again: how to actually check if it hit you
The measurement method for August's spam update, still the right one.
Read the pieceSEO AND GEOGoogle is testing paying publishers. AI overview tracking still beats waiting
How to track AI Overview effects so you do not blame the wrong update.
Read the pieceSEO AND GEOWhen seo reporting tools go dark, the fox reports the gap
Why rank trackers are less reliable exactly when you need them.
Read the piecePaid socialFacebook Marketplace AI: when Meta's Muse agent speaks for the seller
Today's paid social desk: what happens when an agent negotiates for you.
Read the pieceWorried a template or two just caught Google's eye?
folkfox audits AI-assisted content by page type, separates spam-update movement from AI Overview effects, and rebuilds thin pages into ones worth ranking.
Want folkfox in your Google results and AI answers? Set folkfox as a preferred source.

