Claude SEO: How to Get Cited After Anthropic's New Rules

While the marketing world discusses using Claude as a productivity tool, a far more important shift is happening. Anthropic, Claude's creator, has rewritten the rules of the game. Because the company's new policy explicitly bans manipulating AI sources, the only sustainable way to get your brand cited by Claude is to become a primary source of original research and data. The old SEO playbook is not just ineffective; it is now officially discouraged.

This is not a subtle change. It is a clear signal for brands wondering how to appear in AI answers. The tactical race to 'do SEO' on large language models is over before it began. The strategic race to become a citable authority has just begun.

The Old ‘AI SEO’ Playbook Is Now Officially Banned

Effective November 12, 2026, Anthropic's updated usage policy contains a clause that every marketing leader needs to read. As reported by Search Engine Journal on October 9, 2026, the policy now prohibits using its services 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."

This language directly targets the cottage industry of low-effort tactics aimed at gaming AI answers. Think content spun for keywords, articles with fake author personas, or networks of sites designed to look like independent sources. Anthropic is drawing a line in the sand, separating genuine, authoritative information from the noise. For brands, this means any strategy built on shortcuts is now on borrowed time.

The implication is clear: trying to trick Claude is a losing strategy. The only path forward is to create content so valuable and unique that the model has a good reason to cite it. This is a fundamental shift from traditional SEO, where technical signals and keyword density could often compensate for mediocre content. For Claude, the content's inherent authority is paramount.

How Claude Chooses Its Sources (and Why It’s Not Google)

To get cited by Claude, you must first understand how it thinks. Unlike a traditional search engine, Claude is not just crawling the web in real-time to answer every query. It operates on a combination of its vast training data, which has a knowledge cut-off date, and a selective web-browsing capability for fresher information. This makes its sourcing behaviour distinct and, at times, unpredictable.

This sourcing model is also constantly evolving. On October 6, 2026, Anthropic released Claude Sonnet 5.5, a model reported to be over 30% faster and cheaper, which now powers the free tier on its website, as detailed in a model update roundup. Faster, cheaper models mean more answers are generated, increasing the chances for brands to be cited, but also increasing the volatility of those citations. The ground is constantly shifting underfoot.

Claude is also embedding itself into workflows where professionals seek answers. Its recent integration into Google Workspace, announced in early October 2026, allows it to read and edit open documents directly within Google Docs and Sheets, according to a report from Unite.AI. When an AI is inside the document, its recommendations carry even more weight. Getting cited is not just about appearing in a chat window; it is about becoming part of the user's creative process.

This unpredictability is a core feature of all generative AI. According to research from SparkToro, the same AI query changes its answer around 70% of the time. The brand recommendation your first customer sees today may not be what the next one sees tomorrow. This is why spot-checking your AI visibility is a flawed strategy and systematic tracking is essential.

Furthermore, what works for Google may not work for Claude. A study from Ahrefs found that AI visibility correlates more strongly with third-party mentions and video content than with traditional on-page SEO factors. Claude is looking for signals of authority that exist beyond your own website, which is why becoming a source that others reference is so critical.

The Playbook for Earning Citations: Become a Primary Source

If the old playbook is dead, what does the new one look like? It is focused on a single goal: transforming your brand from just another voice on the web into a primary source of information. A primary source creates knowledge; a secondary source merely reports it. Claude, like other advanced AI, is being trained to prefer the former.

Want to get recommended by AI?

Check your AI search visibility, then let the Honeyb agent write, fix, and earn what gets you recommended. Free to start.

Free AI visibility checker

This means shifting your content strategy away from high-volume, low-depth articles and towards creating 'citable assets'. These are pieces of content so original and insightful that other people, and by extension AI models, will reference them. This is less about keywords and more about creating intellectual property.

The table below outlines the strategic shift required. The content on the left is what most brands produce. The content on the right is what gets cited.

The common thread in the right-hand column is verifiable expertise. An original data study presents findings that can be checked. An expert-led analysis carries the weight of the author's reputation. A case study with real numbers provides proof, not just claims. This is the currency of trust for an AI model tasked with providing reliable answers. It cannot evaluate your service directly, but it can evaluate the credibility of the content you publish about it.

This approach aligns with what we already know about how AI models source information. Semrush data shows that Reddit is the single most-cited source in AI answers, accounting for 40.1% of all citations. This is not because Reddit has great on-page SEO; it is because it is a massive repository of firsthand experience and niche discussion. This focus on firsthand experience helps explain why AI models cite Reddit so frequently; it is a vast, unfiltered source of direct opinion.

How to Build Citable Authority: Your First 3 Steps

Shifting to a primary-source strategy can feel daunting, but it can begin with three manageable steps. The goal is not to replace your entire content plan overnight, but to start building the assets that will pay dividends in the age of AI search.

First, audit your existing assets for hidden data. Your company is likely sitting on a trove of unique information. This could be anonymised customer usage data, internal benchmarks, or insights from your sales team's conversations. Packaging this internal knowledge into a public-facing report is often the fastest way to create a citable asset.

Second, plan one piece of keystone research for the next quarter. This does not need to be a massive undertaking. It could be a survey of professionals in your niche, an analysis of public data that no one else has performed, or a definitive white paper that establishes your company's unique perspective on a key industry problem.

Third, track your citations systematically. You cannot improve what you do not measure. Use a dedicated tool to monitor when and where Claude, Gemini, Perplexity, and other engines mention your brand, your competitors, and your key topics. This data will provide the feedback loop needed to understand what is working and refine your strategy over time.

Anthropic's new policy is not a roadblock; it is a map. It points away from the dead end of manipulative tactics and towards the open road of creating genuine value. The brands that follow it will be the ones that AI models, and customers, trust and cite in the years to come.

The first step is to see where you stand. Use our free AI visibility checker to see if Claude and other engines are mentioning your brand today. It is the baseline you need to start building your authority.