Google revealed the official information on Might 15, 2026, particularly addressing how you can optimize for generative AI options in Google Search.
The doc, titled “Optimizing your website for generative AI features on Google Search,” was introduced by John Mueller by means of the Google Search Central Weblog and is now housed below a brand new “Generative AI fundamentals” navigation part in Search Central documentation.
That is Google’s most express, on-record assertion but about what they’re saying works — and what does not — for visibility inside options like AI Overviews and AI Mode.
What the information covers
A lot of the steerage consolidates positions Google has shared at conferences, in weblog posts, and in interviews over the previous yr. It’s organized into 5 important sections:
- Is website positioning nonetheless related for generative AI search?
- Apply foundational website positioning finest practices to generative AI search
- Mythbusting generative AI search: what you needn’t do
- Discover agentic experiences
- Subsequent steps: what to concentrate on
What’s new is that it is now documentation, performing as a reference level for entrepreneurs who’ve been asking for readability.
Google’s core message: AI search visibility remains to be website positioning
The information’s place is evident: website positioning nonetheless issues for generative AI search. Google states clearly that AI Overviews and AI Mode usually are not operating on fully separate methods.
“The most effective practices for website positioning proceed to be related as a result of our generative AI options on Google Search are rooted in our core Search rating and high quality methods,” in keeping with the brand new information.
Google clarifies that its generative AI options use AI strategies like retrieval-augmented generation and query fan-out “to spotlight content material from our Search index.”
In different phrases, in case your content material is not technically sound and high-quality sufficient to rank in conventional search, it will not carry out in AI-generated solutions both.
The mythbusting part: What Google says to cease doing
Google now says SEOs can “ignore” the next techniques for Google Search and its generative AI options:
- llms.txt files: Google’s crawler could uncover these recordsdata, however they’re handled like some other textual content file. There isn’t any particular therapy or most well-liked indexing pathway. (Needless to say different crawlers could make use of such recordsdata.)
- Content chunking: No want to interrupt content material into small items for AI methods, Google says. Google says its methods can perceive multi-topic pages and extract the related passage with out the writer pre-fragmenting the article.
- AI-specific rewriting: AI options can perceive synonyms and normal meanings, in keeping with Google. Rewriting content material to seize each long-tail key phrase variation is not essential, the information says.
- Particular schema or Markdown variations of pages: Not required for Google generative AI search inclusion, the brand new information says.
Google additionally famous that searching for inauthentic “mentions” with a purpose to affect what’s being mentioned about your services isn’t prone to be useful as a result of its generative AI options depend on the identical methods and safeguards because the core rating methods.
“Our core rating methods concentrate on high-quality content material whereas different methods block spam; our generative AI options depend upon each,” in keeping with the information.
The agentic experiences part: A sign for what’s subsequent
Google references agent-friendly finest practices and rising requirements like Universal Commerce Protocol (UCP) and WebMCP, which allow AI brokers to take actions on behalf of customers straight from search outcomes.
The part is framed as non-compulsory and forward-looking.
“If that is one thing that is related to your corporation and you’ve got further time, take a look at the obtainable agentic experiences and evaluate the information to agent-friendly website best practices …” the information says.
Additional studying: Agentic search: How AI agents will decide which brands get found
Why this issues for entrepreneurs
It’s vital to notice that this information applies solely to the Google ecosystem. ChatGPT, Claude, and different AI engines could play by totally different guidelines. And Google could not reveal all the small print about what works and what doesn’t.
Nonetheless, this GEO information is critical for entrepreneurs for 2 causes:
First, the infrastructure round AI search is solidifying, and when infrastructure solidifies, accountability follows. Search entrepreneurs are more and more the professionals finest positioned to guide this work.
Second, the information legitimizes the self-discipline whereas narrowing what that self-discipline truly entails. Google frames AEO and GEO as extensions of website positioning — not separate channels requiring separate experience.
How Semrush helps AI search visibility
Google’s information does not introduce a brand new playbook. Nevertheless it validates one which already works. The query is whether or not your present execution truly holds up. Semrush offers you the information to seek out out.
Use these instruments:
Keyword Magic Tool: Uncover the related key phrases and subjects that create the premise for visibility inside AI-generated solutions.
AI Visibility Toolkit: See which of your URLs are being cited in AI solutions, what prompts are triggering these citations, and the way quotation frequency developments over time throughout platforms.
Enterprise AI Optimization: For enterprise groups — observe AI mentions and citations, analyze sentiment, benchmark towards rivals, determine AI content material gaps, and observe footprint modifications throughout AI system modifications.
Site Audit: Detects whether or not your firewalls block AI crawlers and diagnose dozens of different technical website positioning points..
