What Happened: Google's Spam Policies Now Cover AI Search
On May 15, 2026, Google quietly updated the introduction to its Search spam policies to make one thing explicit: spam now includes “attempting to manipulate generative AI responses in Google Search.” In a single clause, every policy that already governed traditional results — link spam, site reputation abuse, scaled content abuse, cloaking, and more than a dozen others — was confirmed to apply to AI Overviews and AI Mode too. As Search Engine Land reported, Google said the change was made simply “to make it clear that the spam policies apply to all of Google Search, including generative AI responses.”
Alongside the policy edit, Google's guidance on optimizing for generative AI features took direct aim at one tactic in particular. The guide states, in plain language, that “seeking inauthentic ‘mentions’ across the web isn't as helpful as it might seem.” Google's AI features can surface what people say about a brand across blogs, videos, and forum discussions — and a whole cottage industry sprang up promising to manufacture exactly those signals. Google's message: manufacturing them is wasted effort that its ranking and spam systems are built to ignore.
The timing wasn't random. According to industry coverage, the warning followed a well-known AI software platform that began promoting the ability to automate the purchase of brand mentions specifically to land placements inside AI answers across Google, ChatGPT, Claude, and other systems. At Search Central Live events in May 2026, Google's Gary Illyes and Cherry Prommawin addressed the practice head-on. As Search Engine Roundtable documented, they strongly cautioned against buying or manipulating mentions, comparing it to paid links — the kind of signal Google's systems detect, disregard, and ultimately ignore.
Context: This is the enforcement logic of the March 2026 spam update catching up to the AI era. The rules didn't change — Google simply removed any ambiguity about whether they apply when the “result” is an AI-generated answer instead of a ranked list.
What “Inauthentic Mentions” Actually Means
It helps to be precise, because “inauthentic mentions” is guidance language, not a standalone rule. There is no spam policy literally titled “inauthentic mentions.” Instead, the practice of buying or fabricating brand mentions falls neatly under spam policies Google already enforces. Understanding that mapping is the difference between a clear-eyed strategy and an expensive mistake.
In practice, “inauthentic mentions” covers a familiar set of tactics that have simply been rebranded for the AI age:
Paid brand-mention placements
Services that insert your brand into articles, listicles, or “best of” roundups in exchange for payment, dressed up as editorial coverage.
Automated mention generation
Software that mass-creates posts, comments, and references to your brand across the web with the explicit goal of being read by AI crawlers.
Astroturfed community posts
Paid or fake Reddit threads, forum comments, and social posts engineered to look like organic enthusiasm from real users.
Fake or incentivized reviews
Reviews bought, scripted, or exchanged for compensation to inflate the volume of positive sentiment AI systems might pick up.
The reason Google can be relaxed about this is structural. Its generative answers are built on the same Search quality and spam systems that rank traditional results. If those systems are designed to detect and discount paid links and astroturfed signals in the blue links, they carry that same skepticism into the retrieval step that feeds an AI answer. Buying mentions to influence AI is, mechanically, the same bet as buying links to influence rankings — and it now lives under the same spam enforcement umbrella.
Why This Matters for SEO and GEO
Over the past year, “get cited in AI” became its own gold rush. Generative Engine Optimization (GEO) and Answer Engine Optimization (AEO) spawned a wave of vendors, and some of them sell exactly what Google just warned against: packages that promise to plant your brand in the sources AI engines quote. When the conversation shifts from earning citations to buying them, you've crossed from optimization into manipulation — and Google has now drawn that line in writing.
For practitioners, the takeaway is clarity rather than panic. The same brands that built durable visibility in traditional search — by being genuinely useful, citable, and recognizable — are the ones positioned to win in AI search. The shortcut sellers are offering a liability. If you want to understand where legitimate AI citations actually come from, our breakdown of AI citation sources by platform shows how different engines choose what to quote — and none of those mechanisms reward bought mentions.
The vendor red flag
If a GEO/AEO vendor's pitch is “we'll get your brand mentioned in X places so AI picks it up,” ask exactly how those mentions are created. Paid placements, automated posting, and review seeding are the precise behaviors Google described as ineffective and policy-violating. Earned coverage, original data others cite, and real community presence are not.
What the Experts Are Saying
The most useful expert framing came from DemandSphere's Ray Grieselhuber, who summarized Google's guidance with a phrase that has been making the rounds: “AI search is still search.” His point is that retrieval-augmented AI answers are drawn from the same index as everything else, so the fundamentals haven't been replaced — they've been re-weighted.
“If your content is not earning placement in the index for the relevant intent, it cannot be retrieved, and it cannot be cited.”
That reframes the whole problem. Bought mentions don't earn index placement for your own pages; at best they litter other people's pages with references Google's systems are inclined to distrust. The index, as Grieselhuber puts it, is the prize — and you win it with content quality, not purchase orders.
There's also an enforcement signal worth reading carefully. Commentators across the industry noted a familiar Google pattern: the company rarely issues a public warning about a tactic unless it's already building, or has built, the systems to neutralize it. Writing “this won't help” into official documentation is usually the first visible step, not the last. The paid-links comparison is the tell — that playbook ended with widespread devaluation and manual actions.
The most honest nuance, though, came directly from Google. At Search Central Live, Illyes and Prommawin declined to confirm that even ordinary, authentic mentions provide a direct benefit for AI search, saying they weren't sure how useful standard web mentions really are. Read those two statements together and the strategy clarifies: manufactured mentions are a policy risk with no upside, and chasing mentions of any kind — authentic or not — is the wrong unit of optimization. The right one is being a source worth citing.
What Actually Helps You Get Cited in AI
Google's guidance isn't only a list of what to avoid — it's unusually direct about what works. And notably, it tells you to stop doing some things the GEO crowd insists on: you don't need to create special machine-readable files, AI text files, extra markup, or AI-specific rewrites of your content to appear in generative AI search. The model already understands synonyms and meaning. Here's where to put that effort instead.
Publish genuinely original, non-commodity content
Information, analysis, or data that isn't already everywhere. Recycled summaries of what's been said a hundred times give an AI no reason to cite you over the original. Original perspective is the single most durable citation advantage you can build.
Make the page technically eligible to be indexed
If a page can't be crawled and indexed for a query's intent, it can't be retrieved into an AI answer — full stop. Crawlability, clean rendering, and indexing eligibility are table stakes, not optional polish.
Demonstrate real expertise and first-hand experience
Named authors, lived experience, primary research, and specifics that a generalist can't fake. This is the E-E-A-T layer AI systems lean on when deciding which source to trust in an answer.
Keep a fast, clean page experience
Good Core Web Vitals and a usable layout aren't a citation hack, but a slow, broken page undercuts crawl frequency and the overall quality signal your content sends.
Structure for extraction
Clear headings, direct answers near the top, and tight, self-contained sections make your content easy to lift into an answer. This is legitimate formatting, not manipulation — you're making real value easier to retrieve.
None of this is exotic. It's the same foundation that earns rankings, links, and trust — now pointed at a retrieval system instead of a results page. If you want a structured walkthrough of the legitimate playbook, see our guide on how to appear on ChatGPT, Perplexity, and Claude and Google's own Preferred Sources in AI Mode — both reward earned authority, not bought mentions.
Safe vs Risky: A Quick AI-Visibility Audit
Run your current AI-visibility activities — and any vendor proposal on your desk — through this two-column gut check before you spend another dollar.
Risky — policy exposure
- Paying for brand mentions or “AI placement” packages
- Automated tools that post about you across forums and social
- Seeding Reddit threads or reviews that look organic but aren't
- Buying links or sponsored “editorial” inserts for citation bait
- Mass-generating thin pages to repeat your brand and keywords
Safe — earned and durable
- Publishing original research and data others want to cite
- Earning genuine press and coverage through real newsworthy work
- Building authentic community presence and answering real questions
- Encouraging honest customer reviews without scripting or paying
- Improving content depth, structure, and author expertise
If you're unsure whether your off-site footprint leans risky, start by auditing what already points at you. A quick pass with an external link checker surfaces the outbound and inbound references you control, and our roundup of the best link-building tools focuses on earning citations the way Google actually rewards — the same logic that now governs AI mentions.
Tools to Audit Your AI Visibility (the Right Way)
You can't buy your way into AI answers, but you can measure and improve the signals that actually earn citations. These free PikaSEO tools focus on exactly that.
AI Overview Analyzer
Helpful Content Checker
SEO Content Grader
What's Next: Enforcement Is Probably Coming
Right now, the policy is formalized but the detection story for AI-specific mention manipulation is unconfirmed. That gap is exactly where reckless operators will try to squeeze a few more months of bought mentions. It's also exactly the situation that historically precedes a crackdown. With paid links, Google warned, then watched, then devalued and penalized at scale — and the brands caught holding manipulated profiles paid for it.
Expect the same arc here. The most likely developments to watch: clearer examples of “inauthentic mentions” in Google's documentation, signals in Search Console as AI performance reporting matures, and eventually algorithmic devaluation of manufactured mention patterns. None of that requires Google to announce a named “AI mentions update” — devaluation can happen quietly inside the systems that already filter spam.
The strategic move is boring and correct: don't build your AI visibility on a foundation Google has explicitly labeled ineffective and against policy. Invest in being citable. If you're evaluating the broader AI-visibility tooling market while you're at it, our guide to the best AI search visibility tools covers the monitoring stack that helps you track real citation performance — not gamed metrics.
Frequently Asked Questions
About the Author
Co-Founder & SEO Execution
Co-founder of PikaSEO. 11 years in corporate tech, then bootstrapped entrepreneur. Leads SEO execution and content-led growth for SaaS companies.