Google's June Spam Sweep Targets AI Manipulation
Google started rolling out its June 2026 spam update on June 24, and it's going to take a few days to fully settle. The Search Status Dashboard confirmed the timeline, and honestly, that's about all you need to know right now if your rankings are already wobbling.
Here's what makes this one different from the usual spam update cycle: Google explicitly clarified back in May that its spam policies now cover efforts to manipulate generative AI responses in Search. That means buying citations, altering them, or building tactics specifically designed to game AI Overviews and AI Mode all fall under the same rules that have existed for years. The policy wasn't new — just the application of it.
If you've been wondering whether gaming AI answers is a gray area, the answer is no. It's spam. Full stop.
SEO consultant Shushrita M. put it best when she cautioned against panic-reactions during the rollout window: a sudden traffic decline doesn't automatically mean your content is bad. The right move is to identify which page types, queries, and directories got hit, then look for a consistent pattern. Recovery starts with diagnosis, not desperation.
The practical takeaway? Watch your rankings over the next few days before making any changes. Spam updates settle gradually, and premature optimization often makes things worse.
Mueller Explains How AI Impressions Get Counted
John Mueller dropped some clarity on how Google actually measures impressions in the new generative AI report inside Search Console, and there's a nuance most people are missing.
Mueller was responding to Nicola Agius, Director of SEO and Discover at Reach PLC, who asked how impressions work when your content shows up in AI Overviews or AI Mode. The short answer: an impression counts whenever a link to your page appears in those surfaces.
But here's where it gets tricky. Links hidden behind a click-to-expand interaction are only counted when someone actually opens them. So if your content is buried in an expandable section of an AI Overview, you might see low impression numbers even though your page is technically being shown. It's an undercount until the user takes action.
And there's no click data in this report yet. So you're looking at impressions — not clicks, not conversions, just the appearance of your link. That's it.
The implication for publishers and advertisers is worth sitting with: a low impression count doesn't necessarily mean your content isn't being surfaced. It might just mean it's buried under layers of UI friction that users haven't bothered to overcome yet.
Desktop CTR Climbs While Mobile Dips
Advanced Web Ranking published its Q1 2026 CTR benchmark data, and the headline numbers tell a story most people aren't ready for: desktop and mobile are moving in opposite directions.
Mobile's top position dropped roughly 2.2 percentage points. That's not a rounding error — it's a meaningful shift that suggests users on phones are clicking less at the very top of results. Meanwhile, desktop gains showed up mostly below the third position, which is oddly specific and tells you something about where attention is actually landing.
The advice from AWR is straightforward but easy to ignore: review your desktop and mobile CTR separately. Combined figures mask this kind of divergence, and if you're making decisions based on an average, you're probably optimizing for the wrong device.
This isn't a full reversal. It's one quarter of divergence. But the direction matters more than the magnitude when you're allocating budget and adjusting strategy.
Similarweb Shows Where AI Traffic Actually Goes
Similarweb's latest report on AI recommendations reveals something most attribution models completely miss: 55.9% of downstream traffic from ChatGPT recommendations arrives through branded search, not direct AI referrals.
The data comes from a U.S. desktop panel covering finance, travel, and beauty — three sectors where AI recommendations are increasingly influencing purchase decisions. The pattern is consistent: users see a brand mentioned in an AI response, then go search for it by name. They don't click through from the AI interface.
Aleyda Solís, founder of Orainti and a well-known SEO consultant, flagged the measurement blind spot here. She pointed out that AI-influenced demand often arrives through Search and Direct channels, not through anything labeled as an AI referral. Current attribution models have a structural blind spot for this.
The practical implication is significant for anyone tracking AI visibility impact: branded search volume should be treated as a proxy metric. When your brand starts appearing in AI recommendations, the traffic doesn't show up where you'd expect. It shows up as people typing your name into Google.
This also means that if you're only looking at referral traffic from AI platforms, you're probably underestimating the actual reach of your AI visibility by a wide margin.
Google Says SEO Tools Don't Get Special Access
Brendon Kraham, Google's VP of Search and Commerce for Global Ads Solutions, made two statements this week that should shut down a lot of vendor marketing.
First, he said effective SEO aligns with good GEO — meaning the work that drives search visibility also carries into generative experiences. No separate playbook needed.
Second, and more importantly for anyone paying for SEO tools that claim special access: Google doesn't evaluate third-party SEO tools or vendors, and those tools have no access to internal metrics. None.
Cyrus Shepard, who founded Zyppy SEO, agreed with the slogan but pushed back on the reverse direction. He noted that there are a whole lot of things AI-savvy SEOs do right now that they likely would never do if AI had never existed. The symmetry doesn't always hold.
The takeaway is simple: discount any tool or vendor claiming a special pathway into how Google ranks AI answers. The claim has no basis in reality, and Google has said so directly.
The Rules and Meters Get Written Together
This week's theme isn't subtle: Google is writing the rules and the meters at the same time. The spam update expands to cover AI manipulation. Mueller clarifies how impressions are counted — with caveats built in. Kraham confirms no vendor access exists. And independent data from AWR and Similarweb shows what users are actually doing with these surfaces.
The most reliable way to interpret any single number this week is to treat it as provisional. The spam update hasn't settled. The AI impression report lacks click data. Desktop CTR gains don't offset mobile softness. Branded search captures most AI-driven traffic but attribution models can't see it.
Everything is still being measured. Everything is still being defined. And the people who will navigate this best are the ones treating each data point as a signal, not a verdict.