TL;DR

Google and Anthropic are drawing a clearer line between legitimate Generative Engine Optimization and manipulation. Google says its spam policies apply to attempts to manipulate generative AI responses in Search. Anthropic's updated Usage Policy goes further by prohibiting the use of Claude to seed search engines and AI systems with content that misrepresents its origin, authorship, or independence. That does not mean you should stop using AI to create content. It means the safest way to scale is to start with real expertise, disclose relationships, add original value, keep editorial responsibility with a human, and publish content you can stand behind.

We've seen so many spammy tactics over the last 6 months. Clients will call us and say "hey, I just saw someone on TikTok say that if we create lots of websites, it will help us show up in Ai"

Table of Contents

• The line around GEO is getting clearer
• What Anthropic's new policy actually says
• What Google says about spam and AI search
• Why spammy GEO tactics can work temporarily
• How GEO works without the hype
• What crosses the line
• How to scale content without manufacturing authority
• What this means for your brand
• FAQ
• Sources

The line around GEO is getting clearer

Google is no longer the only major technology company pushing back on tactics designed to manipulate AI-generated answers. On October 8, 2026, Anthropic published an updated Usage Policy that takes effect on November 12, 2026. The policy includes a new section called “Do Not Engage in Deceptive Campaigns or Artificial Activity.” It applies to political and commercial activity, including efforts to make coordinated content appear independent. For marketers, the most important part is not a vague warning about bad content. It is a direct statement about influencing the sources used by search engines and AI systems.

Anthropic prohibits using its products to:

“Manipulate the sources from which search engines or AI systems draw answers by seeding them with content that misrepresents its origin, authorship, or independence.”

That language describes a specific kind of GEO spam: creating the appearance of independent agreement when the content actually comes from one coordinated source. SEO and AI search specialist Lily Ray has warned that the industry has seen this pattern before. As quoted by Search Engine Land, Ray put the larger cycle plainly:

“Our industry has gone through these cycles many times before.”

The platforms change. The shortcut changes. The underlying pattern stays familiar.

A tactic finds a weakness in a ranking or retrieval system. Marketers scale it. The web fills with low-value copies. The platform tightens its policies and detection systems. The brands that treated the shortcut like a durable strategy are left with disappearing visibility, cleanup work, and reputation risk.

What Anthropic's new policy actually says

Anthropic is not banning GEO. It is not banning AI-assisted articles. It is not saying that every brand mention, contributed article, comparison page, or third-party placement is deceptive. The policy targets artificial activity and concealed coordination. Examples include fake personas, fabricated media outlets, undisclosed sponsorship intended to influence public opinion, networks of supposedly independent sites that share a common source, and tools built to run deceptive campaigns at scale.

The distinction is important.

You can publish useful content about your business. You can use Claude to organize expertise, develop a draft, repurpose a video, or improve a page. You can contribute to an industry publication. You can earn media coverage. You can work with partners. You can create comparison content when the method, relationship, and evidence are clear. What you should not do is manufacture the appearance of independent authority. If one company creates dozens of sites that look unrelated, gives them invented authors, and uses them to repeat the same recommendation, the reader is being shown a false consensus. If a company pays for endorsements that are presented as neutral editorial judgment, the origin and independence of the recommendation are being concealed.

Anthropic's update turns that distinction into an explicit platform rule for Claude users.

What Google says about spam and AI search

Google has clarified that its existing spam policies apply to generative AI responses in Google Search, including AI Overviews and AI Mode. Google defines spam as techniques used to deceive users or manipulate Search systems into featuring content prominently. Its definition now explicitly includes attempts to manipulate generative AI responses.

That puts several familiar tactics under the same policy umbrella: (YAY!)

Scaled pages created primarily to manipulate rankings or AI answers

• Large amounts of unoriginal content with little value for users
• Doorway pages built around slight query or location variations
• Paid links and placements intended to manipulate ranking signals
• Third-party content that exploits a host site's reputation
• Scraped or lightly rewritten content without meaningful added value
• Hidden content or instructions designed for machines instead of people

Google's guidance does not say AI-generated content is automatically spam. It says using generative AI to produce many pages without adding value may violate the scaled content abuse policy. That is the real boundary: the primary purpose and the value added. If you use AI to help turn real expertise into a clear, useful resource, the tool is assisting the work. If you use AI to flood the web with pages designed mainly to force a ranking, citation, or recommendation, the tool is scaling manipulation.

Why spammy GEO tactics can work temporarily

Spammy GEO tactics became popular for a simple reason: some of them worked. When an AI system searches the web or retrieves documents from an index, repeated claims can influence the material available for an answer. If several pages recommend the same brand, the system may treat that repetition as supporting evidence, especially when it cannot easily see that the pages were coordinated. That created an incentive to manufacture recommendation signals. Some GEO providers began selling high-volume brand mentions, paid listicle placements, fake community conversations, private blog network mentions, and supposedly independent comparison pages. The sales pitch usually focused on citation volume instead of source quality, audience value, or transparent authorship. The short-term result could look convincing. A brand might appear more often in an AI answer. A dashboard might show an increase in mentions. A test prompt might suddenly recommend the client.

But temporary visibility is not the same as durable authority. Search and AI platforms have strong incentives to remove signals that make their answers less useful. Once fabricated recommendations become common enough to reduce answer quality, the platforms can change source weighting, detect coordinated patterns, ignore low-trust pages, penalize sites, restrict accounts, or remove the tactic's value entirely. The risk belongs to the brand even when an outside vendor created the content.

How GEO works without the hype

GEO is the practice of making business information easier for generative systems to understand, retrieve, summarize, compare, and recommend. The process is not identical across every platform or every prompt. Some answers draw primarily from a model's training. Some use a search engine or retrieval index. Some run multiple queries, gather sources, and synthesize a response. Some blend retrieved information with existing model knowledge. The sources cited can also differ from the sources that influenced the wording of the answer.

The practical goal is still straightforward: make accurate, useful information about your brand available in places that people and machines can trust.

That can include:

• A technically healthy, crawlable website
• Clear service and product pages
• Original articles that answer real buyer questions
• Consistent entity information across public profiles
• Structured data that matches visible page content
• First-party research, examples, demonstrations, and case studies
• Earned coverage from relevant publications and communities
• Transparent comparisons with a stated method
• Videos, interviews, and expert commentary that reveal a real point of view
• Strong internal links that make topic and service relationships clear

The goal is not to make the Internet repeat your claim as many times as possible. The goal is to give people and machines enough trustworthy evidence to understand why the claim is true. We've seen our client competitors do spammy, black hat tatics and they DO gain an advantage to those of us trying to do things the right way. It's incredibly frustrating.

What crosses the line

You should treat several patterns as high risk.

Fake independent publishers

Creating or controlling multiple sites that pretend to be unrelated publishers is not third-party validation. It is one source wearing several costumes.

Invented authors or expertise

Fake biographies, fabricated credentials, and false author identities misrepresent where the content came from and why it should be trusted.

Undisclosed paid recommendations

Paid placements can carry advertising, platform, and search-policy obligations. A sponsored recommendation presented as independent editorial judgment creates unnecessary legal and reputational exposure.

Astroturfed community posts

Using fake or aged accounts to pose as customers, start planted conversations, or repeat coordinated recommendations manufactures a consensus that does not exist.

Mass-produced comparison pages

A useful comparison explains the criteria, evidence, tradeoffs, and relationship to the products being reviewed. A thin page that always places the sponsoring brand first is an advertisement pretending to be research.

Content built only for fan-out queries

Creating a separate low-value page for every possible prompt or query variation may run directly into Google's scaled content abuse guidance when the purpose is manipulation rather than user value.

How to scale content without manufacturing authority

You do not have to choose between slow manual publishing and generic AI content. The better option is a human-led content system. Start with a real source of expertise. A short video, customer question, sales call insight, voice note, webinar, product demonstration, or interview can provide the original point of view that a generic prompt cannot. Then give the AI enough context to work accurately. That context can include your services, audience, terminology, examples, boundaries, evidence, and brand position. Use the model to transform that source material into useful formats. One strong video might support an article, email, social posts, short clips, FAQs, or a service-page update. Keep a qualified human responsible for the final content. Review the claims, links, examples, tone, and implications. Remove anything that sounds true but cannot be supported. Make sure the finished piece represents the position your brand is prepared to defend publicly. Capture editorial feedback so the system improves. The goal is not to remove the human. The goal is to stop spending human time on repetitive formatting while preserving judgment, experience, and accountability. This approach creates scale without pretending the work came from somewhere else. It also produces something the Internet needs: more useful expressions of real expertise, not more synthetic agreement. More personal views from people like you and me!

What this means for your brand

The safest GEO strategy is also the strategy most likely to compound. Build your visibility around assets you can keep: a clear website, a well-defined business entity, original expertise, credible relationships, useful content, consistent public information, and proof that supports your claims. Do not measure success only by whether one model mentions your brand this week. Source selection can change quickly. A durable program should also improve search visibility, branded demand, qualified traffic, sales conversations, conversion, and the clarity of your market position.

Before approving a GEO tactic, ask a few direct questions:

• Would a reader understand who created or sponsored this content?
• Does the page add information that does not already exist everywhere else?
• Is the recommendation supported by a clear method or evidence?
• Would the tactic still make sense if search engines and AI systems ignored it?
• Would you be comfortable explaining the tactic publicly to a customer?

If the strategy depends on hiding its origin, authorship, sponsorship, or coordination, it is not a visibility asset. It is a liability waiting for detection to catch up. Google and Anthropic are making that boundary harder to ignore.

FAQ

Is Anthropic banning GEO?

No. Anthropic's updated Usage Policy targets deceptive campaigns and artificial activity, including seeding search and AI sources with content that misrepresents its origin, authorship, or independence. Legitimate content creation and transparent marketing are not prohibited by that language.

Is AI-generated content against Google's rules?

No. Google does not classify content as spam simply because AI helped create it. Google warns that using generative AI to produce many pages without adding value may violate its scaled content abuse policy.

Can you still use Claude to write blog posts?

Yes, provided the use complies with Anthropic's policies. A safe process starts with real expertise, uses AI as an editorial and production tool, and keeps human review and responsibility in place.

Are listicles considered spammy GEO?

Not automatically. A comparison or “best of” article can be useful when it uses a clear method, discloses relevant relationships, includes meaningful evidence, and gives readers an honest view of the tradeoffs. A manufactured list designed only to place a sponsor first carries much more risk.

What is the safest way to improve AI search visibility?

Make your business easier to understand and verify. Publish useful first-party content, strengthen service pages, keep entity information consistent, add appropriate structured data, earn legitimate third-party coverage, and create original material that deserves to be cited.

When does Anthropic's updated policy take effect?

Anthropic published the update on October 8, 2026. The new Usage Policy states an effective date of November 12, 2026.

Related Entities

• Rakos Media Group
• Rakos Media
• Google
• Anthropic
• Claude
• Lily Ray

Related Knowledge

Related Services

• SEO
• AEO
• GEO
• AI Search Optimization
• Knowledge Graph Engineering
• AI Integration and Automation
• Growth Engineering

Related Concepts

• Generative Engine Optimization
• AI Search Visibility
• LLM Visibility
• Entity SEO
• Topical Authority
• Machine-Readable Business Infrastructure
• Structured Data

Related Frameworks

• Search Everywhere Framework™
• Search Everywhere Graph™
• Momentum Framework™

About Rakos Media Group

Rakos Media Group helps businesses improve visibility, trust, conversion, and machine understanding across search, AI systems, websites, paid media, and digital channels. Its work connects SEO, AEO, GEO, AI search optimization, conversion strategy, structured data, and knowledge graph engineering into measurable business infrastructure.

Sources

• Anthropic: 2026 Usage Policy update
• Anthropic: Usage Policy
• Google Search Central: Spam policies for Google web search
• Google Search Central: Google's guide to optimizing for generative AI features on Google Search
• Google Search Central: A new resource for optimizing for generative AI in Google Search
• Search Engine Land: The paid brand mention problem in GEO