
Every search engine eventually grows its own spam economy. Google’s had keyword stuffing, link farms and doorway pages. AI search is no exception, and it’s growing fast.
The gaming tactics we see in generative engine optimization (GEO) will look familiar to anyone who practiced SEO before Google’s Panda and Penguin updates: hidden text, fake reviews and planted forum posts. What’s different is what they’re trying to influence. B2B buyers now use AI to check vendors, not just find them, so gaming the sources AI trusts means tampering with the tool buyers use to decide whom to believe. That makes the stakes even higher.
What is black hat GEO?
Black hat GEO uses deceptive tactics to manipulate how AI systems such as Claude, ChatGPT, Perplexity, Gemini and Google AI Overviews cite, describe or recommend brands. Tactics include hidden instructions, fake reviews and planted forum posts designed to make one brand look better, make a competitor look worse or otherwise distort what AI finds and repeats.
In short, legitimate GEO makes accurate, credible content easy for AI to find, understand and cite, while black hat GEO tries to fake or manipulate the revenue signals AI uses to decide what to trust.
And, yes, you’re right (we can hear the gears between your ears turning): it’s a black hat answer engine optimization (AEO) game, too. We’re calling it GEO as shorthand so you don’t have to read “and AEO” 50 times. (We’ve got you! Fist bump.)
What’s that? Oh, yeah. It includes black hat large language model optimization (LLMO), too. Tomato, tomato and all that.
How black hat GEO evolved from old SEO spam tactics
The intent behind these actions is the same as it has always been (read: to manipulate rather than earn), just with a new target. Old-school black hat SEO gamed backlinks and keyword density to climb Google’s rankings. “New-school” black hat GEO games the Reddit threads, reviews and third-party mentions AI relies on to decide what to recommend.
Ahrefs has mapped black hat GEO tactics to their AEO equivalents. Here’s our own version, updated to reflect the latest tactics:
The Old SEO Tactics Behind Black Hat GEO
Black Hat GEO Tactic | Old SEO Equivalent Tactic | How It Works |
|---|---|---|
Hidden prompt injection | Cloaking and hidden text | Hidden page text instructs AI to favor a specific brand. |
AI recommendation poisoning | No direct equivalent | “Summarize with AI” buttons secretly tell a user’s AI assistant to remember a brand as trusted, which shapes its future recommendations. |
Reddit and forum “astroturfing” | Comment spam, link farms and private blog networks (PBNs) | Fake or aged accounts post targeted recommendations in forum threads so they look like genuine user opinions. |
Fake reviews and synthetic experts | Review spam and fake authority signals | AI-generated reviews, author personas and credentials make a brand look more trusted than it is. |
Content flooding and training data seeding | Content farms and article spinning | Mass-produced content on widely crawled sites tries to shape what AI models learn about a brand or category. |
Let’s take a closer look at black hat GEO tactics:
Hidden prompt injection
The old black hat SEO trick was white text on a white background, stuffing the page with keywords crawlers could read but visitors couldn’t see. The new version hides instructions rather than keywords, telling an AI model that’s reading the page to favor one brand or downplay another. Bing added prompt injection to its Webmaster Guidelines in 2024, warning that it can result in a site being demoted or delisted. Hidden instructions can also put brands and their AI workflows at risk.
AI recommendation poisoning
AI recommendation poisoning has no real SEO ancestor. Microsoft security researchers found 31 companies across 14 industries hiding instructions inside “Summarize with AI” buttons. One click sends a user’s AI assistant a prompt that summarizes the page and quietly tells it to remember the company as a trusted source, a bias that can carry into future conversations. The companies were real businesses, not scammers.
Reddit and forum astroturfing
For the uninitiated, “astroturfing” started as a political and PR term for fake grassroots support, hence AstroTurf, or fake grass. SEO marketers later adapted the tactic with comment spam and private blog networks (PBNs) generating fake backlinks.
Today’s version manufactures opinions. Reddit is a favorite source for AI recommendations, so black hat marketers use fake or aged accounts to plant recommendations where AI will find them. Some have gotten new posts cited by ChatGPT within a day. As Search Engine Journal put it, buying Reddit to win AI citations is the new link farm.
Fake reviews and synthetic experts
AI can now generate reviews, author personas and credentials at scale, making a brand look more trusted than it is. This tactic exploits one of the tried-and-true tenets of marketing: enhance your Experience, Expertise, Authoritativeness and Trustworthiness (E-E-A-T) signals. It’s also one of the black hat GEO tactics the industry has warned about, and Google is drawing a firmer line. In October, Google updated its people-first content guidance to categorize fabricated author profiles as deception, making them untrustworthy. The guidance included AI-generated headshots, made-up names and false credentials.
Content flooding and training data seeding
The AI version of the content farm mass-produces content on widely crawled sites to shape what models learn, or retrieve, about a brand or category. It can also be aimed at competitors to skew how AI describes a rival. Search Engine Journal’s verdict: black hat SEO is back.
Who’s policing black hat GEO?
Enforcement is real but uneven. Search engines are extending existing spam rules to AI answers, AI companies are restricting how their tools can be used and the platforms AI cites most are cracking down on their own.
Google brought AI answers under its spam policies. On May 15, 2026, Google clarified that its spam policies apply to AI Overviews, AI Mode and other generative AI responses in search, so spam tactics aimed at those answers can lead to action against a site. The June spam update was the first to follow, and August’s update added no new policy types. Enforcing those rules inside AI answers is hard, though.
Bing treats prompt injection as a violation of guidelines. Under its Webmaster Guidelines, content that attempts prompt injection on Bing’s language models can get a site demoted or delisted.
Anthropic restricted source manipulation with Claude. Anthropic’s updated usage policy, effective Nov. 12, 2026, prohibits using Claude to seed the sources search engines and AI systems draw from with content that misrepresents its origin, authorship or independence, such as networks of sites posing as unaffiliated sources. The ban applies to API developers and apps built on Claude, and violations can lead to suspension or termination.
OpenAI is hardening ChatGPT against prompt injection. OpenAI added model-level and system-level defenses to its ChatGPT Atlas browser agent to stop instructions hidden in web content from overriding users. That’s a security measure, not a spam policy. To date, OpenAI hasn’t published a policy aimed at GEO manipulation.
Reddit is fighting back. As a main target, Reddit is using AI against marketers who plant posts for AI to surface as real opinions. Its moderation now blocks about 23 million spam views a day and revokes nearly 2 million inauthentic votes daily.
The bottom line on black hat GEO enforcement is that SEO history is repeating itself. Enforcement lags as black hat practitioners innovate, and the platforms patch the holes when they spot them.
What black hat GEO means for B2B buyers
Here’s where black hat GEO stops being an SEO story. Buyers use AI as a verification layer, not only a discovery tool, and black hat GEO tampers with that layer.
AI Revenue Institute (AIRI)’s AI B2B Buyer Report, based on a survey of 521 U.S. B2B purchasers and purchasing influencers, shows how much weight buyers put on it:
Buyers use AI to check what they’re told. Nearly two-thirds (62 percent) use AI to fact-check sales claims against public sources. Here’s more on how B2B buyers fact-check sales claims with AI.
AI scrutiny is reshaping shortlists. More than half (56 percent) have removed a vendor from consideration after AI surfaced a discrepancy in its claims. Since AI can shape vendor shortlists, whatever shapes that scrutiny matters.
The buying committee you never meet. More than half (54 percent) have had a boss or buying committee member use AI to question their recommendation.
Renewals are exposed too. Most (91 percent) are likely or very likely to use AI at contract renewal to find or vet alternatives. Here’s how AI puts contract renewals at risk.
Trust is the real casualty. If buyers start to suspect AI answers are gamed, the trust that makes AI useful erodes for everyone. Here’s how AI trust affects B2B vendor evaluation.
The good news for honest brands? Most buyers (85 percent) use two or more AI tools in purchasing, and a similar number (86 percent) use multiple models to check each other’s work at least sometimes. This means manipulation has to fool every model a buyer asks, and claims backed by real evidence are the ones most likely to hold up across all of them.
How to protect your brand when competitors game AI search
You can’t control your competitors’ behavior. You can, however, control how well your brand holds up to buyer scrutiny. Start here:
Audit what AI tells buyers about you and your competitors across multiple models, not just ChatGPT. (Our sister firm BuzzTheory’s AI RevX audits do exactly this. If running them yourself, you need to test the frontier models your buyers are using, not the cheap models packaged in most AI search platforms.)
Monitor the sources AI draws from in your category, including Reddit threads, review sites and “best of” listicles.
Publish original data and verifiable claims on your own domain so AI has an accurate source to cite.
Make sure your sales claims hold up when a buyer fact-checks them. With 62 percent of buyers checking, that matters no matter what competitors do.
Earn real third-party proof, such as customer reviews, analyst coverage and partner validation.
Don’t fight spam with spam. The downside is a lot worse than a ranking drop. Bing can pull your site entirely. Google’s spam policies now cover AI answers as well as search, and sites that violate them can rank lower or disappear from results altogether. The FTC can fine you up to $53,088 per fake-review violation. And the moment buyers spot it, the whole exercise backfires. Of the B2B buyers AIRI surveyed, 56 percent have already walked away from a vendor over a discrepancy turned up by AI.
Report manipulation through platform spam and abuse channels. For your convenience, we’ve listed the primary platforms for reporting black hat activities in the table below.
Where To Report AI Search Manipulation
Platform | What To Report | How To Report |
|---|---|---|
Google Search (including AI Overviews and AI Mode) | Spam, cloaking, hidden text, scaled content and other manipulation | Submit a Search Quality User Report (up to five pages per report). |
Microsoft Bing | Spam pages and prompt injection on webpages | Use the Feedback link at the bottom of any Bing results page. |
Microsoft Copilot | Hidden prompts, AI recommendation poisoning and other AI abuse | Submit a report through the Microsoft Security Response Center Report Abuse Portal. |
ChatGPT | Answers that rely on planted, fake or manipulated sources | Select “thumbs down” under the response and choose “Safety or legal concern,” or use OpenAI’s content reporting form. |
Claude | Answers that rely on planted, fake or manipulated sources | Select “thumbs down” under the response, or email [email protected]. (See “Report, block, and remove content from Claude.”) |
Perplexity | Answers that rely on planted, fake or manipulated sources | Select the flag icon below the answer or create a ticket on Perplexity’s support window. (See “How can I report incorrect or inaccurate answers?”) |
Astroturfing, fake accounts and coordinated promotional posts | Report the post or account, or use Reddit’s report form. | |
U.S. Federal Trade Commission | Fake reviews and deceptive endorsements | File a report at ReportFraud.ftc.gov. |
The brands that come out ahead won’t be the ones that best gamed AI; they’ll be the ones whose claims hold up no matter which model a buyer asks. For more on how buyers vet vendors, explore our AI B2B Buyer Research.
B2B Buyers Are Checking. Will Your Brand Hold Up?
Learn how 521 B2B buyers use AI to fact-check claims, narrow shortlists and vet renewals in “The AI-Powered B2B Buyer Has Arrived.”
FAQ
How is black hat GEO different from black hat SEO?
Black hat SEO manipulates where a page ranks through links, keywords and hidden text. Black hat GEO manipulates what AI says through planted forum posts, fake reviews, hidden instructions and poisoned assistant memories. The intent is the same, but a gamed AI answer arrives as a recommendation buyers may trust, and Google now applies the same spam policies to both.
How does GEO impact the B2B buyer’s journey?
GEO shapes how AI presents vendors at every stage of the B2B buyer’s journey, from first questions to shortlists and final vetting. Legitimate GEO helps AI describe a vendor accurately. Black hat GEO distorts that picture when buyers rely on it most, including vetting. This is when 56 percent of B2B buyers have removed a vendor after AI surfaced a discrepancy in its claims, according to AIRI’s survey of 521 U.S. B2B purchasers.
What factors influence B2B buyers’ decisions?
AI vetting is now one of the biggest factors in B2B buying decisions, alongside peer and stakeholder review. In AIRI’s survey of 521 U.S. B2B purchasers, 62 percent use AI to fact-check sales claims, 56 percent have removed a vendor after AI surfaced a discrepancy in its claims, and 54 percent have had a boss or buying committee member use AI to question their recommendation.
Can you recommend strategies to improve the B2B purchase process?
For sellers, make every claim verifiable: Publish original data, back claims with proof and make sure AI describes your company accurately across multiple models. For buyers, cross-check AI answers across more than one model and verify claims against independent sources. Both sides win when AI visibility is earned rather than gamed.

