Google just redrew the line between legitimate AI visibility optimization and spam. On May 15, 2026, a single sentence added to Google's Search spam policies made it official: attempting to manipulate generative AI responses in Google Search is now explicitly classified as spam. If you're investing in GEO (Generative Engine Optimization), this changes your risk calculus overnight.

TL;DR

Google updated the opening paragraph of its spam policies to include manipulation of AI Overviews and AI Mode alongside traditional ranking manipulation. The previous definition only covered attempts to rank content highly. The new definition extends to any attempt to manipulate generative AI responses. Tactics like orchestrated Reddit mentions, self-promotional listicles, and citation pyramid schemes now have an explicit policy framework for penalization.

What Exactly Changed

The updated introductory sentence now reads: "spam refers to techniques used to deceive users or manipulate our Search systems into featuring content prominently, such as attempting to manipulate Search systems into ranking content highly or attempting to manipulate generative AI responses in Google Search."

The previous version only referenced ranking content highly. Seven words were added, but the implications run deep. Google's potential action remains the same: lower ranking or complete removal from results, which now explicitly includes exclusion from AI Overviews and AI Mode citations.

Barry Schwartz at Search Engine Land reported the change on the same day, noting that many popular GEO tactics could now violate these policies.

Key Takeaways

AI Overviews and AI Mode are now under the spam policy umbrella. Any tactic designed to artificially boost your brand's presence in AI-generated responses can trigger penalties.
"Recommendation poisoning" has a policy framework. The term describes coordinated efforts to make AI systems recommend specific brands through inauthentic signals.
This pairs with Google's new AI Search optimization guide released the same day, creating an official carrot-and-stick approach: here's how to do it right, and here's what happens if you cheat.
Measurement is not manipulation. Tracking where and how AI engines cite your brand remains legitimate. Artificially inflating those citations is not.

The Decision Framework: White, Grey, and Red

Not every GEO tactic falls neatly into "safe" or "spam." Here's a practical framework for evaluating your current strategy against the updated policy.

White zone (clearly legitimate):

Monitoring AI citations across ChatGPT, Perplexity, Gemini, and AI Overviews
Publishing original research and expert content that AI engines cite naturally
Structured data optimization (JSON-LD for Organization, Product, FAQ) to help AI parse your offerings
Building genuine brand authority through PR, partnerships, and thought leadership
A/B testing content formats to understand what earns organic AI citations

Grey zone (proceed with caution):

Aggressive schema markup beyond what your content actually contains (schema-bait)
Community engagement at scale in Reddit, Quora, or LinkedIn with subtle brand mentions
Competitive citation displacement strategies targeting specific AI prompts

Red zone (now explicitly policy-violation territory):

Orchestrated Reddit mention farming with coordinated accounts recommending your brand
Self-promotional listicles designed to rank your brand #1 in AI-curated "best of" answers
Citation pyramid schemes where satellite sites cross-link to inflate perceived authority
Fake LinkedIn endorsements or review manipulation aimed at AI training data
Inauthentic mention farming across forums, Q&A sites, and social platforms

How to Audit Your GEO Program

If you're working with a GEO vendor or running an in-house AI visibility program, ask these three questions:

"Would this tactic make sense if AI search didn't exist?" Genuine content marketing, PR, and brand building pass this test. Orchestrated mention campaigns do not.
"Are we creating signals or amplifying real ones?" Tracking and measuring your AI visibility is amplifying real signals. Manufacturing fake community discussions is creating artificial ones.
"Could Google trace this back to a coordinated campaign?" If the answer is yes, assume they will.

This is the same logic Google has applied to link building for over a decade. The playbook is familiar: measure and earn, don't manufacture.

What This Means for AI Visibility Strategy

The May 15 update draws a clear boundary. Brands that invested in legitimate AI search optimization through quality content, structured data, and genuine authority are on the right side of the line. Those relying on grey-hat GEO playbooks from the past 12 months face real enforcement risk.

This update also validates the measurement-first approach. Tools that track AI citations and monitor brand presence across AI engines serve a fundamentally different purpose than tools designed to manipulate those citations. The former is analytics. The latter is now officially spam.

The message from Google is straightforward: build something worth recommending, then measure how AI engines respond. Don't try to game the recommendation.

FAQ

Does this policy change affect all AI search engines or just Google?

The May 15 update applies specifically to Google's AI Overviews and AI Mode. However, other AI search engines like Perplexity and ChatGPT have their own quality guidelines. Tactics that violate Google's policy likely conflict with other platforms' terms as well.

Can Google actually detect AI citation manipulation?

Google has over two decades of experience detecting coordinated inauthentic behavior in traditional search. The same pattern-detection capabilities apply to identifying manufactured signals designed to influence AI responses. Coordinated Reddit threads, listicle farms, and review pyramids leave detectable footprints.

Is structured data optimization still safe after this update?

Yes. Adding accurate schema markup that reflects your actual content is explicitly encouraged in Google's companion AI Search optimization guide. Schema-bait, where markup claims capabilities or reviews that don't exist, crosses the line.

What should I do if my GEO vendor uses grey-hat tactics?

Audit their deliverables against the red zone list above. Ask for transparency on their methods. If they rely on manufactured mentions, coordinated community campaigns, or citation pyramid schemes, those tactics now carry explicit policy risk.

Sources

Google Search spam policies (updated May 15, 2026)
Search Engine Land: Google confirms spam policies apply to AI Overviews and AI Mode by Barry Schwartz
Search Engine Roundtable: Google Says Spam Policies Apply To AI Responses In Search
Gizmodo: Google's Spam Policies Now Apply to Attempts to Manipulate AI

Google's Spam Policy Now Covers AI Overviews: Is Your GEO Strategy at Risk?