Google released the August 2026 Google spam update at about 12:30 pm ET on August 18. The rollout finished three days later, at 4:50 am ET on August 21. Google called it a normal spam update and named no new policy, so most SEO teams filed it under rankings and moved on. That reading missed half the story. Three months earlier, Google had written AI Overviews and AI Mode directly into its spam policies. The bar that decides whether a page ranks now also decides whether it can be cited.
Key Takeaways
- The August 2026 spam update kept Google’s policies the same and improved automated detection of them. Pages that used to sit just under the detection line are now more likely to be caught.
- Since May 15, 2026, Google’s spam policies explicitly cover generative AI responses in Search. That means a spam demotion also limits a page’s eligibility for AI Overviews and AI Mode.
- Many tactics sold as AI optimisation, such as content at scale, separate pages for AI crawlers and bought mentions, match existing spam policies. They are a liability for citation.
What Did Google Actually Change on August 18?
On paper, nothing. Search Engine Roundtable quoted Google’s own description: “This is a normal spam update, and it will roll out for all languages and locations.” No link-spam target named. No breakdown of affected query share. The policy text itself didn’t move a word.
The threshold moved anyway. Google’s documentation frames a spam update as a notable improvement to its automated detection systems, chiefly SpamBrain, its AI-based spam-prevention system. The written rules stay fixed. The classifier just gets sharper at applying them. A page that cleared the bar in July, simply because the system hadn’t yet learned to spot its pattern, can fail in August under that exact same rule.
Pace is the other tell. August was the third spam update of 2026, following rollouts on March 24 and June 24. Three in five months. Google also warns that recovery can take many months, since its systems need time to confirm a site has actually changed.
The May Rewrite That Decides What a Spam Update Touches
On May 15, 2026, Google edited the opening definition in its spam policies. Spam used to mean techniques that manipulate Search systems into ranking content highly. The new wording tacks on one more phrase: “attempting to manipulate generative AI responses in Google Search.”
PPC Land reported the scope plainly: the edit added no rules, it just applied the existing ones to new surfaces. Every policy that once covered blue links, scaled content abuse, cloaking, site reputation abuse, inauthentic mentions, now covers AI Overviews and AI Mode by name. Google’s own documentation says its generative AI features depend on both its core ranking systems and its spam systems.
So August was the second spam update to land after the rewrite. June was the first. Each improvement to SpamBrain now hits two surfaces at once.
Why Eligibility Comes Before Citation
AI Overviews build their answers from indexed web content. Per PPC Land’s reading of Google’s enforcement history, the spam systems act on ranking signals before any individual search feature gets involved. A page that’s demoted or dropped is removed at that early stage. Everything downstream, including the AI answer, picks from what’s left.
Citation is a contest among eligible pages, and Google’s spam systems decide which pages are eligible before any AI answer is assembled.
That fits the two-layer model of AEO. Training-data presence builds slowly through wide coverage, and a single spam update barely dents it. Retrieval presence is the layer that decides what actually gets cited in a live answer, and it can vanish inside a three-day rollout. Other engines run their own retrieval and quality filters, and Google’s classifier has no say over those. But the patterns SpamBrain hunts for, volume without differentiation, stitched-together content, are exactly the patterns any retrieval system has reason to discount too.
Share of Voice and Share of Citation can split apart here. A brand hit by the update keeps its mentions scattered across the web, so its Share of Voice holds steady. Its own pages drop out of the candidate pool, so its Share of Citation falls. The AI answer might still name the brand, just while citing a competitor or a third-party review as the source.
unique Moburst pages cited in AI answers, part of 42,435 total AI citations and a number 1 position in category.See the case study.
Citation volume at that scale comes from many pages, and each one has to earn its own eligibility. Lose it across a whole site section and the damage compounds: every page that drops out also loses the citations it would have attracted next.
Which Pages Sit Closest to the Line?
Scaled content abuse comes first. The policy defines it as generating many pages “for the primary purpose of manipulating search rankings and not helping users,” and it names generative AI tools as one way to do it. It also covers stitching together content from several pages without adding value. Templated location pages, programmatic comparisons, AI-written explainers that just repackage the top ten results: all of it fits that definition.
Inauthentic mentions is risk area number two. As early as May 2025, practitioners were already showing that self-promotional “best of” listicles, including ones published on a brand’s own site, were getting cited inside AI Overviews. Google’s documentation now says chasing those mentions is less useful than it looks. Paid placements and reciprocal mention deals carry the same exposure.
Two more tactics, both popular in AI optimisation circles, sit just as close to the line. PPC Land reported that Google and Bing signalled in February 2026 that keeping separate markdown pages, or altered content, just for AI crawlers counts as cloaking. In January 2026, Danny Sullivan warned against chopping content into bite-sized chunks for LLMs. Structured data and clean content architecture on the canonical page? Fine. A second, hidden version of that page built for machines? That’s the risk.
What to Do Before the Next Refresh
Google runs periodic refreshes of each spam update. Treat August as a deadline that hasn’t closed yet.
- Break Search Console impressions down by directory, then compare August 18-21 against the two weeks before. A drop confined to one templated section points to pattern-level classification, not a sitewide problem.
- Go through every page you produced at scale. For each one, write down what it offers that its source pages don’t. Can’t answer that? Consolidate it or pull it from the index, which is the exact step Google’s own policy asks for.
- Pull any page version served only to AI crawlers.
- Rewrite or retire listicles that rank your own brand first. Keeping one anyway? Publish the criteria and disclose the interest.
- Review brand mention programmes for paid or reciprocal placements before a classifier finds them for you.
- Track AI Overview citations page by page, every week, next to rankings. A page that stops getting cited while its competitors keep showing up is often the first sign it’s lost eligibility.
Plan on months, not weeks. Google says its systems need time to learn that a site complies, so expect any fix to surface gradually, across several refresh cycles.
FAQs
Did the August 2026 Spam Update Introduce New Content Quality Rules?
No. Google described it as a normal spam update that applies globally and to all languages. It improved automated detection of existing spam policies, so pages that previously went undetected can now be caught under the same written rules.
Do Google’s Spam Policies Apply to AI Overviews and AI Mode?
Yes. On May 15, 2026, Google updated its spam policies to state that spam includes attempting to manipulate generative AI responses in Google Search. Every existing policy now applies to AI Overviews and AI Mode.
Can a Page Hit by a Spam Update Still Be Cited in AI Overviews?
It becomes much less likely. Google’s spam systems act on ranking signals before individual features, so a demoted or removed page drops out of the pool AI answers draw from. Third-party pages that mention the brand can still be cited.
How Long Does Recovery Take After a Spam Update?
Google says recovery can take many months. Its automated systems need time to learn that a site complies with the spam policies, and Google runs periodic refreshes during which improvements can show up.