# Free GEO & AEO course with certificate (AYKA Academy)

Learn how ChatGPT, Gemini, Perplexity and Google's AI features find, read and recommend businesses, and what to change on your own website. 13 tracks, 65 short lessons with their sources, a final exam and a PDF certificate you can add to LinkedIn. Free for everyone: no account, no email, nothing to pay.

Course page: https://getayka.com/academy/
Exam and certificate: https://getayka.com/academy/exam/
Publisher: AYKA Technologies (https://getayka.com)
Research status: 2026-11-09

## 1. How AI finds you
Memory, live search, and why every answer is different.
Page: https://getayka.com/academy/how-ai-finds-you/

### 1.1 Memory or live search? (Documented)

- An AI assistant answers from two places: what it learned in training (its memory) and a live web search it runs for your question.
- Each company uses different bots for the two jobs. OpenAI: GPTBot collects training data, OAI-SearchBot feeds ChatGPT search. Anthropic: ClaudeBot and Claude-SearchBot. Google: Google-Extended for Gemini training, Googlebot for Search and its AI features.
- So you have two jobs: be readable and indexed today (technical), and be talked about widely enough to end up in the memory (reputation).

Do it on your site: Write down 10 questions a real visitor asks before choosing you, for example “opening hours [museum] Sunday”. You will reuse them in track 10.

Sources: OpenAI: Overview of OpenAI crawlers <https://developers.openai.com/api/docs/bots>; Google Search Central: Optimizing your website for generative AI features <https://developers.google.com/search/docs/fundamentals/ai-optimization-guide>

### 1.2 AI Overviews and AI Mode (Documented)

- Google's AI features first retrieve pages from the Google Search index, then write the answer. Google calls this grounding (retrieval-augmented generation).
- AI Mode splits your question into many related searches at once, a “query fan-out”. A question about weeds in a lawn also searches “best herbicides for lawns” and “remove weeds without chemicals”.
- AI Overviews appear only when Google thinks they add something, so they often don't show at all. A page must be indexed and allowed a snippet to be a supporting link.

Do it on your site: Search your 3 most important questions on Google. Note whether an AI Overview appears and which sites it links.

Sources: Google Search Central: Optimizing your website for generative AI features <https://developers.google.com/search/docs/fundamentals/ai-optimization-guide>; Google Search Central: AI features and your website <https://developers.google.com/search/docs/appearance/ai-features>; Google: AI Mode update (I/O 2025) <https://blog.google/products-and-platforms/products/search/google-search-ai-mode-update/>

### 1.3 Whose index? (Documented)

- Not every assistant uses Google. Microsoft Copilot cites from Bing: Bing's AI Performance report counts citations across Copilot and Bing's AI summaries.
- ChatGPT, Claude and Perplexity run their own search crawlers: OAI-SearchBot, Claude-SearchBot, PerplexityBot. OpenAI does not say publicly which other indexes it uses, so claims that ChatGPT is “Bing-powered” are not proven.
- Practical rule: be indexed in Google and Bing, and let the assistants' search bots in.

Do it on your site: Search site:yourdomain on Google and on Bing. If key pages are missing in one of them, fix indexing first.

Sources: OpenAI: Overview of OpenAI crawlers <https://developers.openai.com/api/docs/bots>; Anthropic Help Center: Does Anthropic crawl data from the web? <https://support.anthropic.com/en/articles/8896518>; Perplexity: Perplexity crawlers <https://docs.perplexity.ai/docs/resources/perplexity-crawlers>; Bing Webmaster blog: AI Performance in Bing Webmaster Tools (public preview) <https://blogs.bing.com/webmaster/February-2026/Introducing-AI-Performance-in-Bing-Webmaster-Tools-Public-Preview>

### 1.4 Why answers differ (Evidence-based)

- AI answers change from run to run. In a January 2026 study, SparkToro and Gumshoe had volunteers run the same prompts thousands of times in ChatGPT, Claude and Google AI.
- Getting the same list of brands twice was very rare, and the same order rarer still. The stable signal was how often a brand appears. (Reported by Search Engine Journal; not peer-reviewed.)
- AI Mode is also personalised: it can use your earlier searches. Two people asking the same question may see different sources.

Do it on your site: Ask one of your questions three times in a fresh chat. Count in how many answers you appear.

Sources: Google Search Help: Personal Intelligence in AI Mode <https://support.google.com/websearch/answer/16011537>; Search Engine Journal: AI recommendations change with nearly every query (SparkToro) <https://www.searchenginejournal.com/ai-recommendations-change-with-nearly-every-query-sparktoro/566242/>

### 1.5 Clicks in an AI world (Evidence-based)

- Pew Research (July 2025, 900 US adults, 68,879 searches): with an AI summary, people clicked a normal result in 8% of visits, without one in 15%. They clicked a link inside the summary in 1%.
- Ahrefs found the click rate of the top result about 34.5% lower when an AI Overview appeared (March 2025 data), and about 58% lower in its February 2026 update. These are correlations, not proof.
- Seer Interactive saw brands cited inside the AI Overview get more clicks than brands not cited. Measure conversions and enquiries, not only traffic.

Do it on your site: In your analytics, pick one goal that matters (booking, enquiry, donation, application) and check it monthly next to traffic.

Sources: Pew Research Center: Google users are less likely to click when an AI summary appears <https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/>; Ahrefs: AI Overviews reduce clicks <https://ahrefs.com/blog/ai-overviews-reduce-clicks/>; Ahrefs: AI Overviews reduce clicks (update) <https://ahrefs.com/blog/ai-overviews-reduce-clicks-update>; Seer Interactive: AIO Impact on Google CTR, 2026 Update (53 brands, 5.47M queries) <https://www.seerinteractive.com/insights/aio-impact-on-google-ctr-2026-update>

## 2. Open the door
robots.txt, AI bots, CDNs and Google's controls.
Page: https://getayka.com/academy/open-the-door/

### 2.1 robots.txt in 10 minutes (Documented)

- robots.txt is a public text file at yourdomain/robots.txt. It tells crawlers what they may read. It is a voluntary standard (RFC 9309), not a lock.
- Each block starts with “User-agent:” (the bot's name) and then “Disallow:” or “Allow:” rules. A bot that is not named follows the “User-agent: *” block.
- Add a “Sitemap:” line so crawlers find all your pages.

Do it on your site: Open yourdomain/robots.txt. Check that Googlebot, Bingbot, OAI-SearchBot, Claude-SearchBot and PerplexityBot are not blocked, and that a Sitemap line exists.

Sources: IETF RFC 9309: Robots Exclusion Protocol <https://www.rfc-editor.org/rfc/rfc9309>; Google Search Central: AI features and your website <https://developers.google.com/search/docs/appearance/ai-features>

### 2.2 The AI bot zoo (Documented)

- Training bots collect text for future models: GPTBot, ClaudeBot, Google-Extended, Applebot-Extended, CCBot.
- Search bots index pages for AI answers: OAI-SearchBot, Claude-SearchBot, PerplexityBot. Blocking Claude-SearchBot may reduce your visibility in Claude's search results, says Anthropic.
- User bots fetch a page when a person asks: ChatGPT-User, Claude-User, Perplexity-User. OpenAI says robots.txt may not apply to ChatGPT-User; Perplexity-User generally ignores it; Anthropic says Claude-User respects it.

Do it on your site: Play the bot game in this Academy until you score 8/8.

Sources: OpenAI: Overview of OpenAI crawlers <https://developers.openai.com/api/docs/bots>; Anthropic Help Center: Does Anthropic crawl data from the web? <https://support.anthropic.com/en/articles/8896518>; Perplexity: Perplexity crawlers <https://docs.perplexity.ai/docs/resources/perplexity-crawlers>; Search Engine Journal: Anthropic's Claude bots make robots.txt decisions more granular (via) <https://www.searchenginejournal.com/anthropics-claude-bots-make-robots-txt-decisions-more-granular/568253/>

### 2.3 Block training, keep search (Documented)

- You can say no to training and still be found. Name the training bots with “Disallow: /” and the search bots with “Allow: /”.
- Blocking Google-Extended does not remove you from AI Overviews: those are controlled through Googlebot and snippet rules.
- Whether to block training is your decision. There is no measured evidence either way for small organisations: record it as a conscious choice.

Do it on your site: Copy the template from the Agent's fix pack (or this lesson's source), adapt the sitemap URL, and ask whoever runs your site to publish it.

Sources: OpenAI: Overview of OpenAI crawlers <https://developers.openai.com/api/docs/bots>; Search Engine Journal: Anthropic's Claude bots make robots.txt decisions more granular (via) <https://www.searchenginejournal.com/anthropics-claude-bots-make-robots-txt-decisions-more-granular/568253/>; Google Search Central: AI features and your website <https://developers.google.com/search/docs/appearance/ai-features>

### 2.4 Your CDN may block AI (Documented)

- A perfect robots.txt is useless if your hosting or CDN blocks the bot first. Google lists checking “any CDN or hosting infrastructure” as a best practice.
- Since 1 July 2025, Cloudflare blocks AI crawlers by default on new domains and offers Pay Per Crawl.
- Since 15 September 2026, Cloudflare's “Block” and “Block on pages with ads” settings also stop mixed-use crawlers, including Googlebot, Bingbot and Applebot, so they cut you out of search too. To refuse training but stay in search, use the new “Disallow AI Training” setting.

Do it on your site: Ask your web host or check your Cloudflare dashboard: which AI crawlers are blocked? Allow the search bots you want.

Sources: Cloudflare: Content Independence Day <https://blog.cloudflare.com/content-independence-day-no-ai-crawl-without-compensation/>; TechCrunch: Cloudflare's new policy pushes AI companies to pay <https://techcrunch.com/2026/07/01/cloudflares-new-policy-pushes-ai-companies-to-pay-for-publishers-content/>; Cloudflare: Accountable mixed-use AI crawlers <https://blog.cloudflare.com/accountable-mixed-use-ai-crawlers>; Google Search Central: AI features and your website <https://developers.google.com/search/docs/appearance/ai-features>

### 2.5 Snippets and Google's AI opt-out (Documented)

- To limit what Google shows from a page, use nosnippet, data-nosnippet (for parts of a page), max-snippet or noindex.
- Since 31 August 2026, every site has a Search Console setting called “Search generative AI”: you can exclude your site from AI Overviews, AI Mode and AI features in Discover.
- It does not cover the Gemini app or AI training: that is still Google-Extended. Sites that opt out get no traffic or impressions from these features; Google says it is not a ranking signal elsewhere.

Do it on your site: If one paragraph should never be quoted (for example outdated prices), wrap it in data-nosnippet instead of hiding the whole page.

Sources: Google Search Central: AI features and your website <https://developers.google.com/search/docs/appearance/ai-features>; Google: New opportunities, control and insights for website owners <https://blog.google/products-and-platforms/products/search/new-controls-website-owners/>; PPC Land: Google lists 7 ways to keep content out of Search AI (via) <https://ppc.land/google-lists-7-ways-to-keep-content-out-of-search-ai-opt-out-included/>

## 3. Be readable
Text, JavaScript, sitemaps, llms.txt and languages.
Page: https://getayka.com/academy/be-readable/

### 3.1 Text beats images and PDFs (Documented)

- Google asks that important content is available as text. Opening hours in a picture, prices in a PDF, dates only in a poster: machines may miss them.
- Images and videos still count: Google says that following its image and video SEO guides already prepares them for AI search.
- Add alt text to images and transcripts to videos. They help people with disabilities and machines.

Do it on your site: Find one fact that only exists in an image or PDF (hours, prices, deadlines) and put it on the page as plain text.

Sources: Google Search Central: AI features and your website <https://developers.google.com/search/docs/appearance/ai-features>; Google Search Central: Optimizing your website for generative AI features <https://developers.google.com/search/docs/fundamentals/ai-optimization-guide>

### 3.2 JavaScript and AI crawlers (Open debate)

- Google renders JavaScript. Most AI companies do not say whether their bots do. Perplexity, for example, does not document it.
- Safe assumption: facts that only appear after JavaScript runs in the browser may be invisible to some AI bots.
- Sada's Audit compares the raw HTML with the rendered page and warns you when most text is missing without JavaScript.

Do it on your site: Run Sada's Audit on your home page and look at “Content without JavaScript”.

Sources: Google Search Central: AI features and your website <https://developers.google.com/search/docs/appearance/ai-features>; Klarivo: robots.txt for AI crawlers, quoting Perplexity docs (via) <https://www.klarivo.ai/blog/robots-txt-for-ai-crawlers>

### 3.3 Sitemaps and IndexNow (Documented)

- An XML sitemap lists your pages and is Google's documented way to help discovery. Submit it in Search Console and Bing Webmaster Tools.
- IndexNow lets you tell Bing and some other engines instantly that a page changed. Google does not use it.
- Most website systems (WordPress plugins, Shopify, TYPO3 extensions) create both for you: switch them on.

Do it on your site: Open yourdomain/sitemap.xml. If it is missing, switch on your system's sitemap feature and submit it in both webmaster tools.

Sources: Google Search Central: AI features and your website <https://developers.google.com/search/docs/appearance/ai-features>; IndexNow <https://www.indexnow.org/>

### 3.4 llms.txt: the honest story (Open debate)

- llms.txt is a proposed file that summarises a site for AI. Google lists it among unnecessary tactics: Google Search ignores it, it neither helps nor harms.
- SE Ranking (Nov 2025, ~300,000 domains) found it on about 10% of sites and no effect on how often AI cites them. One log study found only ~0.1% of AI-bot requests touched it.
- Other side: some practitioners argue that assistants other than Google may read it, but no AI provider documents using it. Cheap to add, but no magic fix.

Do it on your site: If you have llms.txt, keep it short and accurate. Spend your time on robots.txt and content first.

Sources: Google Search Central: Optimizing your website for generative AI features <https://developers.google.com/search/docs/fundamentals/ai-optimization-guide>; SE Ranking: llms.txt study (~300,000 domains) <https://seranking.com/blog/llms-txt/>; Search Engine Journal: Google's AI search guide calls AEO and GEO 'still SEO' <https://www.searchenginejournal.com/googles-new-ai-search-guide-calls-aeo-and-geo-still-seo/575026/>

### 3.5 Many languages, right to left (Documented)

- If you publish in several languages, link the versions with hreflang so search engines show the right one.
- Each page needs the correct lang attribute. Arabic pages also need dir="rtl" so text and punctuation display correctly.
- Translate properly: a machine-translated page that nobody checked makes you look less trustworthy in every language.

Do it on your site: Run Sada's Audit on one page per language and fix any “language” or “hreflang” warning.

Sources: Google: Localized versions of your pages (hreflang) <https://developers.google.com/search/docs/specialty/international/localized-versions>; W3C: Structural markup and right-to-left text in HTML <https://www.w3.org/International/questions/qa-html-dir>

## 4. Be recognised
One name everywhere, schema, Wikidata.
Page: https://getayka.com/academy/be-recognised/

### 4.1 One name everywhere (Documented)

- AI has to decide that “Stadt Albstadt”, “Stadtverwaltung Albstadt” and albstadt.de are the same organisation. Identical facts make that easy.
- Use the same name, address, phone, opening hours and website on your site, Google Business Profile, Bing Places, Apple Business Connect, Facebook and LinkedIn.
- Google lists keeping Business Profile and Merchant Center information up to date as a best practice for its AI features.

Do it on your site: Make a table: platform, name, address, phone, hours, link. Fill it for 5 platforms and correct every difference.

Sources: Google Search Central: AI features and your website <https://developers.google.com/search/docs/appearance/ai-features>

### 4.2 Organization schema and sameAs (Documented)

- Structured data (JSON-LD) states facts in a machine format: name, logo, address, website. sameAs lists your official profiles elsewhere.
- Google says no special schema is needed for its AI features, but structured data must match the visible text.
- Google still documents Organization, Local business, Event, Profile page, Article and Video structured data.

Do it on your site: Use the code in Sada's fix pack: one Organization block with name, url, logo, address and sameAs. Check it with Google's Rich Results Test.

Sources: Google Search Central: AI features and your website <https://developers.google.com/search/docs/appearance/ai-features>; schema.org <https://schema.org>; Google: Structured data search gallery <https://developers.google.com/search/docs/appearance/structured-data/search-gallery>

### 4.3 The right type (Documented)

- schema.org has specific types: Museum, CollegeOrUniversity, School, NGO, GovernmentOrganization, LocalBusiness and many more.
- Pick the most specific type that is true. A memorial site is not a generic Organization; a town hall is a GovernmentOrganization.
- Specific types help machines tell you apart from organisations with similar names.

Do it on your site: Look up your type on schema.org and replace a generic “Organization” with the specific one.

Sources: schema.org <https://schema.org>

### 4.4 Wikidata before Wikipedia (Documented)

- Wikipedia needs significant coverage in reliable, independent sources, and has conflict-of-interest rules when you write about yourself.
- Wikidata has a lower bar: a clearly identifiable entity that can be described with serious, public references.
- A correct Wikidata item (official website, location, founding date, identifiers) is a clean fact sheet many systems can read.

Do it on your site: Search Wikidata for your organisation. Correct the item, or create one with official website, location and founding date, each with a reference.

Sources: Wikidata: Notability <https://www.wikidata.org/wiki/Wikidata:Notability>; Wikipedia: Notability <https://en.wikipedia.org/wiki/Wikipedia:Notability>

### 4.5 What Google retired (Documented)

- FAQ rich results stopped appearing in Google Search on 7 May 2026, and Google removed their documentation in June 2026. HowTo rich results are gone too.
- Many other features are still documented (November 2026): Organization, Local business, Event, Course list, Education Q&A, Profile page, Article, Video and Review snippet. Dataset markup now only serves Dataset Search.
- You don't need to delete FAQ markup, and visible FAQ text still helps readers. Just don't expect dropdowns in Google.

Do it on your site: If a provider sells you FAQ schema for “rich results”, ask them to read Google's May 2026 notice.

Sources: Inblog: Google FAQ rich results deprecation, quoting Google's notice (via) <https://inblog.ai/blog/google-faq-schema-rich-result-deprecation>; Google: Structured data search gallery <https://developers.google.com/search/docs/appearance/structured-data/search-gallery>; Google Search Central: Latest documentation updates <https://developers.google.com/search/updates>

## 5. Write to be quoted
First-hand content, answers first, real evidence.
Page: https://getayka.com/academy/write-to-be-quoted/

### 5.1 Non-commodity content (Documented)

- Google's 2026 guide: content people find unique, compelling and useful will likely influence your presence in AI search more than any other tip.
- Commodity content repeats what everyone says (“7 tips for visiting a museum”). Non-commodity content only you can write (“What we found when we restored our 1928 organ”).
- Your archives, your numbers, your people and your local knowledge are things no AI can copy from elsewhere.

Do it on your site: List 3 things only your organisation knows or has. Turn one into a page this month.

Sources: Google Search Central: Optimizing your website for generative AI features <https://developers.google.com/search/docs/fundamentals/ai-optimization-guide>

### 5.2 Answer first (Open debate)

- Start pages with the direct answer: what, for whom, where, with the key fact. Then explain.
- Agencies report more citations with answer-first pages; no AI company documents it. Google says special “chunking” for AI is unnecessary.
- It costs nothing and helps readers, so do it anyway, with headings and short paragraphs.

Do it on your site: Rewrite the first two sentences of your three key pages so they answer the main question directly, with a date or number.

Sources: Google Search Central: Optimizing your website for generative AI features <https://developers.google.com/search/docs/fundamentals/ai-optimization-guide>; Aggarwal et al.: GEO: Generative Engine Optimization (KDD 2024) <https://arxiv.org/abs/2311.09735>

### 5.3 Sources, quotes, numbers (Evidence-based)

- The GEO study (Princeton, Georgia Tech, AI2, IIT Delhi; KDD 2024) found that adding citations, quotations and statistics raised a source's visibility in AI answers by up to 40%.
- Keyword stuffing did worse than doing nothing.
- Limits: the main tests used a simulated engine on 2023-era models, and “visibility” meant share of the answer text, not clicks. Treat it as a direction, not a law.

Do it on your site: Add one named source, one quotation or one dated figure to your most important page.

Sources: Aggarwal et al.: GEO: Generative Engine Optimization (KDD 2024) <https://arxiv.org/abs/2311.09735>

### 5.4 Publish your own data (Documented)

- A figure only you have is citeable: visitors per year, students placed in jobs, objects in the archive, trees planted.
- Date it, say how it was counted, and put it in text (not only in a chart).
- This is exactly the first-hand material Google's 2026 guide puts first.

Do it on your site: Publish one dated, attributed figure: “In 2025, 41,200 people visited the memorial (ticket count).”

Sources: Google Search Central: Optimizing your website for generative AI features <https://developers.google.com/search/docs/fundamentals/ai-optimization-guide>

### 5.5 Don't overdo it (Documented)

- Making a separate page for every variation of a question, mainly to influence AI answers, breaks Google's scaled content abuse policy.
- Mass-produced AI pages, hidden text and fake freshness dates are spam, not optimisation.
- One excellent page that answers a topic fully beats twenty thin ones.

Do it on your site: Find pages that say almost the same thing. Merge them into one and redirect the rest.

Sources: Google Search Central: Optimizing your website for generative AI features <https://developers.google.com/search/docs/fundamentals/ai-optimization-guide>; Google: Spam policies for Google web search <https://developers.google.com/search/docs/essentials/spam-policies>

## 6. Reputation beyond your site
Where AI looks, reviews, press, communities.
Page: https://getayka.com/academy/reputation-beyond-your-site/

### 6.1 Where AI looks (Evidence-based)

- Peec AI analysed 30 million cited sources across ChatGPT, AI Mode, Gemini, Perplexity and AI Overviews: Reddit, YouTube, LinkedIn, Wikipedia and Forbes led.
- Assistants differ: Profound found Wikipedia far more often in ChatGPT, Reddit more in Perplexity.
- Shares swing fast and studies disagree: Ahrefs measured Reddit falling from 19.1% to 2.3% of US ChatGPT answers (March–September 2026) and Wikipedia from 14.1% to 2.9%, while BrightEdge reports Reddit growing in Google's AI Overviews. Spread your presence.

Do it on your site: Ask an assistant one of your questions and note which websites it cites. Those are your target places.

Sources: Search Engine Land: AI search engines cite Reddit, YouTube and LinkedIn most (Peec AI) <https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138>; Profound: AI platform citation patterns <https://www.tryprofound.com/blog/ai-platform-citation-patterns>; Ahrefs: ChatGPT citation study, 66.6M responses (title only, URL not verified); BrightEdge: Reddit in AI Overviews (title only, URL not verified); Semrush: Most cited domains in AI <https://www.semrush.com/blog/most-cited-domains-ai/>; Ross Simmonds: ChatGPT Reddit citations <https://rosssimmonds.com/blog/chatgpt-reddit-citations/>

### 6.2 Reviews the honest way (Documented)

- Reviews and business data feed Google's AI through your Business Profile.
- Fake or bought reviews are illegal: the US FTC rule since 21 October 2024 (including AI-written reviews), the EU Omnibus Directive since 28 May 2022.
- Ask real visitors, students or partners, with no reward tied to a positive rating.

Do it on your site: Ask five genuine recent visitors or partners for an honest Google review this week.

Sources: FTC: Rule on the Use of Consumer Reviews and Testimonials (16 CFR 465) <https://www.ftc.gov/legal-library/browse/rules/rule-use-consumer-reviews-testimonials>; EU Omnibus Directive (EU) 2019/2161 <https://eur-lex.europa.eu/eli/dir/2019/2161/oj>

### 6.3 Directories and lists (Evidence-based)

- Directories, tourism portals, sector associations and “best of” lists are where AI often finds recommendations; review sites such as Yelp appeared often in recommendation queries.
- A wrong entry spreads wrong facts. A missing entry leaves you out of the list.
- Check the ones assistants actually cite for your questions first.

Do it on your site: List 10 directories, portals or registers in your sector. Correct or create your entry in three of them.

Sources: Search Engine Land: AI search engines cite Reddit, YouTube and LinkedIn most (Peec AI) <https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138>

### 6.4 Communities without spam (Open debate)

- Forums like Reddit are heavily cited, so marketers try to plant posts. That breaks platform rules and can break consumer law.
- What works ethically: answer real questions as yourself, say who you are, link only when it helps.
- Whether community activity changes AI answers for small organisations is not measured. Do it to help people.

Do it on your site: Find one real question about your field in a forum and answer it helpfully, with your role stated.

Sources: Search Engine Land: AI search engines cite Reddit, YouTube and LinkedIn most (Peec AI) <https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138>; Google: Spam policies for Google web search <https://developers.google.com/search/docs/essentials/spam-policies>

### 6.5 Press and partners (Documented)

- Independent coverage is what Wikipedia calls “significant coverage in reliable sources” and what assistants trust.
- Partners (universities, cities, foundations, funders) can link and describe you accurately.
- Give journalists and partners one fact sheet: name, one-line description, numbers with dates, contact, logo.

Do it on your site: Write a one-page fact sheet and send it to two partners asking them to check how they describe you.

Sources: Wikipedia: Notability <https://en.wikipedia.org/wiki/Wikipedia:Notability>

## 7. Social channels
Which platforms AI can read, and how to set up yours.
Page: https://getayka.com/academy/social-channels/

### 7.1 Which platforms let AI in (Documented)

- LinkedIn's robots.txt (checked 9 Nov 2026) blocks all bots it does not name. GPTBot, ClaudeBot, PerplexityBot and CCBot are not named, so they are blocked.
- YouTube's robots.txt names no AI bot and blocks only paths like /results or /api/: watch pages are open.
- Sada's Channel check reads each platform's robots.txt live, because these files change.

Do it on your site: Open your channels with Sada's panel open and read the “AI and search crawlers” result for each.

Sources: LinkedIn robots.txt (checked 9 Nov 2026) <https://www.linkedin.com/robots.txt>; YouTube robots.txt (checked 9 Nov 2026) <https://www.youtube.com/robots.txt>; Meta: Web crawlers documentation <https://developers.facebook.com/documentation/sharing/webmasters/web-crawlers>

### 7.2 Instagram is searchable (Documented)

- Since 10 July 2025, public posts of professional accounts (users 18+) can be indexed by Google and Bing.
- Personal and private accounts are excluded. You can switch it off in privacy settings.
- So captions now matter beyond Instagram: write what, who, where and when in plain words.

Do it on your site: Set your account to Professional and Public, and check that “Allow public photos and videos to appear in search engine results” is on.

Sources: SEOZoom: Instagram posts indexed by Google (via) <https://www.seozoom.com/instagram-google-indexing/>

### 7.3 YouTube descriptions and captions (Evidence-based)

- YouTube is among the most cited sources in AI answers (Peec AI, 2026).
- A machine reads the title, the description and the captions, not the pictures.
- Auto-captions often get names wrong, especially German, French and Arabic names.

Do it on your site: Add 2–3 factual sentences to the descriptions of your 5 most-watched videos and correct their captions.

Sources: Search Engine Land: AI search engines cite Reddit, YouTube and LinkedIn most (Peec AI) <https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138>; YouTube robots.txt (checked 9 Nov 2026) <https://www.youtube.com/robots.txt>

### 7.4 LinkedIn company pages (Open debate)

- LinkedIn is among the most cited domains, even though most AI crawlers are blocked. How the content gets there (search indexes, partnerships, user fetches) is not documented.
- Fill in the About tab: description, website, industry, location. Sada reads it on your computer to compare with your site.
- Use the same name and description as on your website.

Do it on your site: Open your LinkedIn page with Sada and fix every “Improve” item.

Sources: LinkedIn robots.txt (checked 9 Nov 2026) <https://www.linkedin.com/robots.txt>; Search Engine Land: AI search engines cite Reddit, YouTube and LinkedIn most (Peec AI) <https://searchengineland.com/ai-search-engines-cite-reddit-youtube-and-linkedin-most-study-473138>

### 7.5 Bios and link-in-bio (Open debate)

- A bio is a one-line fact sheet: name, what you are, where, your own website.
- Link-in-bio pages add a third-party layer between the channel and your site. No study measures the effect, but your own domain is the clearest signal.
- Your website should link back to each channel, ideally in its sameAs markup.

Do it on your site: Copy Sada's suggested bio, check the facts, and use your own domain as the main link.

Sources: Google Search Central: AI features and your website <https://developers.google.com/search/docs/appearance/ai-features>

## 8. Local and sector playbooks
Museums, universities, NGOs, awards, town halls.
Page: https://getayka.com/academy/local-and-sector-playbooks/

### 8.1 Business Profile for AI (Documented)

- Google's own Business Profile and Maps data feed its AI answers. Google lists keeping it up to date as a best practice.
- Complete everything: category, hours, special hours, website, photos, description, attributes such as accessibility.
- Answer reviews, and update the profile the same day your hours change.

Do it on your site: Open your Google Business Profile and fill in every empty field. Add special hours for the next holidays.

Sources: Google Search Central: AI features and your website <https://developers.google.com/search/docs/appearance/ai-features>; Google Search Central blog: A new resource for optimizing for generative AI <https://developers.google.com/search/blog/2026/05/a-new-resource-for-optimizing>

### 8.2 Museums and memorial sites (Documented)

- Put hours, closures, prices, accessibility and how to get there on one page as text, identical to your Business Profile.
- Mark up exhibitions and events with Event structured data (name, dates, place).
- Accessibility information is also required in many cases by EU accessibility law and helps every visitor.

Do it on your site: Create or fix one “Plan your visit” page with all practical facts in text.

Sources: Google: Structured data search gallery <https://developers.google.com/search/docs/appearance/structured-data/search-gallery>; Directive (EU) 2019/882 (European Accessibility Act) <https://eur-lex.europa.eu/eli/dir/2019/882/oj>

### 8.3 Universities and schools (Documented)

- One page per programme: language of teaching, duration, fees, dated deadlines, requirements for international applicants.
- Google still documents Course list markup, but says its AI features need no special schema. Clear, complete programme pages matter most.
- Label or redirect past deadlines: an AI that quotes last year's date sends students away.

Do it on your site: Check your top programme page: are language, fees and this year's deadline written as text?

Sources: Google: Structured data search gallery <https://developers.google.com/search/docs/appearance/structured-data/search-gallery>; Google Search Central: AI features and your website <https://developers.google.com/search/docs/appearance/ai-features>

### 8.4 NGOs, awards and town halls (Documented)

- NGOs and foundations: a transparency page with registration number, annual report and leadership.
- Awards and competitions: a permanent page per edition with call, deadline, jury and dated winners. Never overwrite past winners.
- Town halls: service pages titled as citizens' questions, with hours and documents as text. In Germany the Impressum now falls under § 5 DDG.

Do it on your site: Pick the rule for your sector and fix one page today.

Sources: Google Search Central: Optimizing your website for generative AI features <https://developers.google.com/search/docs/fundamentals/ai-optimization-guide>; § 5 Digitale-Dienste-Gesetz (Impressum) <https://www.gesetze-im-internet.de/ddg/__5.html>

### 8.5 Sensitive history (Open debate)

- No AI company publishes how it ranks sources on contested history such as the Nazi era or memorial culture.
- Google's rater guidelines treat topics that affect society and civic understanding as needing high trust (YMYL).
- What you control: name sources and archives, date your statements, say who wrote them and why, give context. Whether assistants favour official institutions is not proven.

Do it on your site: Add author, date and sources to one historical page.

Sources: Google Search Quality Rater Guidelines (YMYL) <https://guidelines.raterhub.com/searchqualityevaluatorguidelines.pdf>

## 9. AI agents
Assistants that book, buy and fill in forms.
Page: https://getayka.com/academy/ai-agents/

### 9.1 What agents do (Documented)

- AI agents act for people: compare, book, buy. Google announced “agentic checkout” in AI Mode in May 2025.
- In 2026 Google presented the Universal Commerce Protocol with retailers; OpenAI and Stripe built the Agentic Commerce Protocol, and OpenAI later shifted toward merchants' own checkouts.
- For most organisations the near-term job is simpler: make facts and forms usable.

Do it on your site: Write down the one action you want people to take online (book, register, apply, donate).

Sources: Google: AI Mode update (I/O 2025) <https://blog.google/products-and-platforms/products/search/google-search-ai-mode-update/>; Agentic commerce protocol tracker <https://www.agentic-commerce.sh/en/protocols/>

### 9.2 Agent-friendly forms (Open debate)

- Labelled fields, real buttons and clear error messages are WCAG 2.2 basics for people with disabilities.
- The same clarity likely helps browser agents too, though that link is not measured.
- CAPTCHAs and logins stop agents; Anthropic says its bots don't bypass CAPTCHAs.

Do it on your site: Run your booking or donation form through a free accessibility checker (for example WAVE) and fix missing labels.

Sources: W3C: WCAG 2.2 <https://www.w3.org/TR/WCAG22/>

### 9.3 Protocols explained (Documented)

- MCP (Model Context Protocol, Anthropic, Nov 2024) connects AI tools to data and services.
- WebMCP lets a web page offer functions as tools to an agent in the browser: an early Chrome preview in Feb 2026, a W3C Community Group draft in Oct 2026, not shipped by other browsers.
- NLWeb (Microsoft) is early-stage. UCP and ACP are commerce protocols mainly for shops.

Do it on your site: None of these is required today for non-shops. Note them, and re-check next year.

Sources: Agentic commerce protocol tracker <https://www.agentic-commerce.sh/en/protocols/>; Advanced Web Ranking: WebMCP and UCP <https://www.advancedwebranking.com/blog/webmcp-ucp-architecture-of-agentic-web>

### 9.4 Facts before login (Documented)

- An agent or crawler cannot see what is behind a login.
- Show prices, dates, availability and conditions as text before anyone signs in.
- This also helps people compare you before they commit.

Do it on your site: Check your booking path: is the price visible before login? If not, add it to the public page.

Sources: Google Search Central: Optimizing your website for generative AI features <https://developers.google.com/search/docs/fundamentals/ai-optimization-guide>

### 9.5 What to ignore for now (Documented)

- You don't need MCP, WebMCP or commerce protocols to be found and recommended in 2026.
- Beware of offers to “make your site agent-ready” with custom protocol builds if your basics (crawl access, text, forms) are not done.
- Revisit when your platform (WordPress, Shopify, booking tool) supports it out of the box.

Do it on your site: Ask any vendor proposing protocol work which browser or assistant uses it today, with a source.

Sources: Agentic commerce protocol tracker <https://www.agentic-commerce.sh/en/protocols/>; Advanced Web Ranking: WebMCP and UCP <https://www.advancedwebranking.com/blog/webmcp-ucp-architecture-of-agentic-web>

## 10. Measure
Free reports, referrals, your question sheet, logs.
Page: https://getayka.com/academy/measure/

### 10.1 Search Console's Generative AI report (Documented)

- Since June 2026 (worldwide since 31 August 2026), Search Console has Generative AI reports for Search and Discover: impressions, pages, countries, devices, dates.
- Limits: impressions only, with no clicks and no search queries; data starts around 18 May 2026. Smaller sites may not see the report yet.
- AI data also stays inside the normal Performance report.

Do it on your site: Open Search Console → Performance → Generative AI. Note impressions and your top 5 pages.

Sources: Google Search Central blog: Generative AI performance reports <https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports>; Search Engine Journal: Search Console AI reports rolled out worldwide <https://www.searchenginejournal.com/google-search-console-ai-reports-rolled-out-worldwide/587836/>; Google: New opportunities, control and insights for website owners <https://blog.google/products-and-platforms/products/search/new-controls-website-owners/>

### 10.2 Bing AI Performance (Documented)

- Bing Webmaster Tools (public preview since Feb 2026) shows citations across Copilot, Bing AI summaries and some partners.
- It also shows “grounding queries”: the search phrases the AI used to find your pages. A goldmine for wording.
- No clicks either. A June 2026 update reportedly added citation share and topics.

Do it on your site: Verify your site in Bing Webmaster Tools and read your top grounding queries.

Sources: Search Engine Land: Bing Webmaster Tools AI Performance report (via) <https://searchengineland.com/bing-webmaster-tools-ai-performance-report-468751>; Bing Webmaster blog: AI Performance in Bing Webmaster Tools (public preview) <https://blogs.bing.com/webmaster/February-2026/Introducing-AI-Performance-in-Bing-Webmaster-Tools-Public-Preview>

### 10.3 AI referrals in analytics (Documented)

- ChatGPT search adds utm_source=chatgpt.com to its links.
- Many AI clicks arrive without a source, so analytics undercounts them.
- Build one filter for sources containing chatgpt, perplexity, copilot, gemini, claude. Sada's fix pack contains a ready regex.

Do it on your site: In GA4, create an exploration filtered on those sources and save it.

Sources: OpenAI: Overview of OpenAI crawlers <https://developers.openai.com/api/docs/bots>

### 10.4 Your question sheet (Evidence-based)

- Agencies measure AI visibility with a fixed set of 20–50 real questions, asked several times per assistant, every month.
- The key metric is the share of runs in which you appear, not a “position”.
- Note also the competitor named, the cited URL and any wrong fact about you.

Do it on your site: Use Sada's Ask AI tab: pick 10 questions, run them in 3 assistants, export the CSV monthly.

Sources: Search Engine Journal: AI recommendations change with nearly every query (SparkToro) <https://www.searchenginejournal.com/ai-recommendations-change-with-nearly-every-query-sparktoro/566242/>

### 10.5 Server logs for bots (Documented)

- Your server logs are your own proof that AI bots visit.
- Anyone can fake a user agent. OpenAI and Perplexity publish IP lists to check real bots.
- Shorten visitors' IP addresses to respect privacy: you only need the bots.

Do it on your site: Ask your host for one week of logs and count visits from OAI-SearchBot, PerplexityBot, Claude-SearchBot and Googlebot.

Sources: OpenAI: Overview of OpenAI crawlers <https://developers.openai.com/api/docs/bots>; Klarivo: robots.txt for AI crawlers, quoting Perplexity docs (via) <https://www.klarivo.ai/blog/robots-txt-for-ai-crawlers>

## 11. Myths and your monthly routine
Spot false promises and run your own plan.
Page: https://getayka.com/academy/myths-and-monthly-routine/

### 11.1 Ten GEO myths (Documented)

- Google, May 2026: optimising for generative AI search “is still SEO”. You can ignore llms.txt, chunking, AI-specific rewriting and special schema for Google.
- Keyword stuffing hurt in the GEO study. llms.txt showed no effect on citations in a 300,000-domain study.
- Real levers: crawl access, indexing, first-hand content, consistent facts, independent mentions.

Do it on your site: Play the myth game until you score 8/8.

Sources: Search Engine Journal: Google's AI search guide calls AEO and GEO 'still SEO' <https://www.searchenginejournal.com/googles-new-ai-search-guide-calls-aeo-and-geo-still-seo/575026/>; Aggarwal et al.: GEO: Generative Engine Optimization (KDD 2024) <https://arxiv.org/abs/2311.09735>; SE Ranking: llms.txt study (~300,000 domains) <https://seranking.com/blog/llms-txt/>

### 11.2 Vetting an agency (Documented)

- A “guaranteed #1 in ChatGPT” cannot be verified: answers change nearly every run.
- Hidden text, prompt injection and mass AI pages break Google's spam policies.
- Ask three questions: How many runs per question do you measure? Which official documentation supports your tactics? Do you use hidden text, fake reviews or mass pages?

Do it on your site: Send these three questions to any provider before you sign.

Sources: Google Search Central: Optimizing your website for generative AI features <https://developers.google.com/search/docs/fundamentals/ai-optimization-guide>; Search Engine Journal: AI recommendations change with nearly every query (SparkToro) <https://www.searchenginejournal.com/ai-recommendations-change-with-nearly-every-query-sparktoro/566242/>; Google: Spam policies for Google web search <https://developers.google.com/search/docs/essentials/spam-policies>

### 11.3 Your one-page monthly report (Documented)

- Report six lines: AI impressions (Search Console), AI citations (Bing), AI referral visits and goals (analytics), appearance rate per assistant (your sheet), wrong facts fixed, next three actions.
- Keep it to one page so leadership reads it.
- Compare month to month, not day to day: AI answers vary.

Do it on your site: Make the template now and fill it on the first Monday of each month.

Sources: Google Search Central blog: Generative AI performance reports <https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports>; Search Engine Land: Bing Webmaster Tools AI Performance report (via) <https://searchengineland.com/bing-webmaster-tools-ai-performance-report-468751>; Search Engine Journal: AI recommendations change with nearly every query (SparkToro) <https://www.searchenginejournal.com/ai-recommendations-change-with-nearly-every-query-sparktoro/566242/>

### 11.4 Refresh calendar (Documented)

- Quarterly: your top 10 pages (deadlines, prices, hours, contacts).
- Yearly: About page, leadership, statistics, annual report, Wikidata item.
- After every change: update website, Business Profile and channels on the same day.

Do it on your site: Put four quarterly reminders in your calendar called “Sada refresh”.

Sources: Google Search Central: Optimizing your website for generative AI features <https://developers.google.com/search/docs/fundamentals/ai-optimization-guide>

### 11.5 Impact versus effort (Documented)

- Quick wins with high impact: robots.txt and CDN settings, consistent Business Profile, answer-first intros, one Organization schema block.
- Bigger projects: original data, independent coverage, multilingual quality.
- Do quick wins this month, plan one bigger project per quarter.

Do it on your site: Run Sada's Agent and tick off the first three plan items marked “you”.

Sources: Google Search Central: AI features and your website <https://developers.google.com/search/docs/appearance/ai-features>

## 12. Law and ethics (EU)
Training opt-outs, AI Act, privacy, accessibility, honesty.
Page: https://getayka.com/academy/law-and-ethics-eu/

### 12.1 Opting out of AI training (Documented)

- EU law (DSM Directive, Art. 4) allows text and data mining unless the rights holder reserves it in an appropriate way, such as machine-readable means for online content.
- In Kneschke v. LAION (OLG Hamburg, 5 U 104/24, 10 Dec 2025), a plain-language opt-out from 2021 was not shown to be machine-readable, and the burden of proof lay with the rights holder. Whether plain language can count today was left open.
- The case is not final: Germany's Federal Court of Justice (BGH, I ZR 281/25) heard it on 3 September 2026 and may refer it to the EU Court of Justice. Safest today: name the training bots in robots.txt. Not legal advice.

Do it on your site: If you want to opt out, add the training-bot block to robots.txt and note the decision.

Sources: Directive (EU) 2019/790 (DSM), Art. 3 and 4 <https://eur-lex.europa.eu/eli/dir/2019/790/oj>; Open Future: LAION round 2 (OLG Hamburg, 5 U 104/24) <https://openfuture.eu/blog/laion-round-2-machine-readable-but-still-not-actionable/>; Inside Tech Law: Machine-readable opt-outs and AI training, Hamburg court clarifies <https://www.insidetechlaw.com/blog/2025/12/machine-readable-opt-outs-and-ai-training-hamburg-court-clarifies-copyright-exceptions>; BFF: Kneschke ./. LAION e.V., what the BGH hearing showed <https://bff.de/news/kneschke-laion-e-v-was-die-verhandlung-vor-dem-bgh-gezeigt-hat/>

### 12.2 AI Act and the GPAI Code (Documented)

- The EU AI Act (Art. 53) requires providers of general-purpose AI models to have a policy to respect EU copyright, including Art. 4 opt-outs.
- The voluntary GPAI Code of Practice (July 2025) commits signatories to respect machine-readable opt-outs such as robots.txt, as legal commentators report.
- The Digital Omnibus (Regulation (EU) 2026/1744, in force since 27 July 2026) moved deadlines: high-risk obligations under Annex III now apply from 2 December 2027. Research and heritage institutions keep their own TDM exception (Art. 3).

Do it on your site: If you are a museum or research body, note that Art. 3 lets you mine texts for research too.

Sources: Directive (EU) 2019/790 (DSM), Art. 3 and 4 <https://eur-lex.europa.eu/eli/dir/2019/790/oj>; Usercentrics: EU AI Act high-risk delay (Digital Omnibus) (via) <https://usercentrics.com/knowledge-hub/eu-ai-act-high-risk-delay-article-50-transparency-consent/>

### 12.3 Privacy-safe measurement (Documented)

- Analytics cookies generally need prior consent (Germany: § 25 TDDDG; France: CNIL guidance). Server logs can contain personal data under GDPR.
- Load analytics only after consent and mention it in your privacy notice.
- For bot analysis, shorten human IP addresses: you only need bot traffic.

Do it on your site: Check that your analytics loads only after consent.

Sources: Directive (EU) 2019/790 (DSM), Art. 3 and 4 <https://eur-lex.europa.eu/eli/dir/2019/790/oj>

### 12.4 Accessibility law as an ally (Documented)

- The European Accessibility Act applies since 28 June 2025 to specified services such as e-commerce, banking and ticketing (Germany: BFSG). Micro-enterprises providing services are exempt; public bodies have their own directive.
- Accessible pages have text alternatives, labels and clear structure.
- Whether that also makes sites easier for AI to read is plausible but not proven.

Do it on your site: Run one key page through an accessibility checker and fix the top three issues.

Sources: Directive (EU) 2019/882 (European Accessibility Act) <https://eur-lex.europa.eu/eli/dir/2019/882/oj>

### 12.5 Impressum and honest marketing (Documented)

- In Germany, provider details (Impressum) fall under § 5 DDG since May 2024. A complete Impressum also gives one consistent source for your legal name and address.
- Fake reviews are banned in the EU and the US. Disguised advertising breaks consumer law.
- Honesty is also good GEO: assistants repeat what is consistent and verifiable.

Do it on your site: Check that your Impressum is complete and matches your Business Profile.

Sources: § 5 Digitale-Dienste-Gesetz (Impressum) <https://www.gesetze-im-internet.de/ddg/__5.html>; FTC: Rule on the Use of Consumer Reviews and Testimonials (16 CFR 465) <https://www.ftc.gov/legal-library/browse/rules/rule-use-consumer-reviews-testimonials>; EU Omnibus Directive (EU) 2019/2161 <https://eur-lex.europa.eu/eli/dir/2019/2161/oj>

## 13. New in November 2026
Second research pass: what changed, and what we corrected.
Page: https://getayka.com/academy/whats-new/

### 13.1 One shared AI-training opt-out (Documented)

- On 30 October 2026, Google, OpenAI, Microsoft and Anthropic backed one shared robots.txt token for AI training: “User-agent: __usage_ai-training”, based on the IETF AIPREF work.
- It is not live yet: Google aims for early 2027, Microsoft from 1 February 2027, OpenAI “in the coming months”. Until then, keep naming each training bot.
- Careful: Google reads this token like Google-Extended, which also covers grounding in the Gemini app. It does not control AI Overviews or AI Mode: that is the Search Console setting.

Do it on your site: Put a reminder in your calendar for February 2027: check the Google and Bing crawler docs and decide whether to add the shared token.

Sources: Google Search Central blog: A shared control for AI training <https://developers.google.com/search/blog/2026/10/ai-training-control>; PPC Land: Google, Microsoft, OpenAI and Anthropic back one AI-training opt-out (via) <https://ppc.land/google-microsoft-openai-and-anthropic-back-one-ai-training-opt-out/>; Google Search Central: Latest documentation updates <https://developers.google.com/search/updates>

### 13.2 Cloudflare: “Block” now hits search (Documented)

- Since 15 September 2026, Cloudflare's “Block” and “Block on pages with ads” settings apply to mixed-use crawlers, including Googlebot, Bingbot and Applebot.
- So switching on Block can quietly remove you from Google and Bing search, not only from AI training.
- To refuse training and stay in search, Cloudflare added “Disallow AI Training”, and marks Apple, Google and Microsoft as “Accountable” crawlers that remain allowed for search.

Do it on your site: If you use Cloudflare, open the AI crawler settings and check that “Block” is off unless you mean it. Use “Disallow AI Training” to refuse training only.

Sources: Cloudflare: Accountable mixed-use AI crawlers <https://blog.cloudflare.com/accountable-mixed-use-ai-crawlers>; TechCrunch: Cloudflare's new policy pushes AI companies to pay <https://techcrunch.com/2026/07/01/cloudflares-new-policy-pushes-ai-companies-to-pay-for-publishers-content/>

### 13.3 robots.txt in court (Documented)

- USA: in the OpenAI copyright litigation (S.D.N.Y., 15 December 2025), Judge Stein ruled that robots.txt is not an access control under the DMCA, comparing it to a “keep off the grass” sign.
- EU: a machine-readable reservation matters for the text-and-data-mining exception (OLG Hamburg, December 2025). Germany's Federal Court of Justice heard the appeal on 3 September 2026; no decision yet.
- Takeaway: technically robots.txt is a request, but in the EU it can be your legal reservation. Use it deliberately and keep a dated record.

Do it on your site: Save a dated copy of your robots.txt every time you change it.

Sources: Eric Goldman: Are robots.txt instructions legally binding? (Ziff Davis v. OpenAI) <https://blog.ericgoldman.org/archives/2025/12/are-robots-txt-instructions-legally-binding-ziff-davis-v-openai.htm>; Inside Tech Law: Machine-readable opt-outs and AI training, Hamburg court clarifies <https://www.insidetechlaw.com/blog/2025/12/machine-readable-opt-outs-and-ai-training-hamburg-court-clarifies-copyright-exceptions>; BFF: Kneschke ./. LAION e.V., what the BGH hearing showed <https://bff.de/news/kneschke-laion-e-v-was-die-verhandlung-vor-dem-bgh-gezeigt-hat/>; Open Future: LAION round 2 (OLG Hamburg, 5 U 104/24) <https://openfuture.eu/blog/laion-round-2-machine-readable-but-still-not-actionable/>

### 13.4 North Africa and the Gulf (Documented)

- Saudi Arabia's new Copyright Law (in force 12 August 2026) explicitly allows copying lawfully published works to develop AI, without permission or payment (Art. 26(4)). Republishing or directly exploiting them stays forbidden.
- Data protection: the Saudi PDPL is fully enforced since September 2024. The UAE PDPL is in force, but its Executive Regulations had not been issued by mid-2026. Tunisia: Organic Law 2004-63 still applies; reform bill 095/2025 is in committee, not adopted.
- No country-specific fake-review rule was found in Tunisia, Morocco, Egypt, the UAE or Saudi Arabia, but general misleading-advertising and consumer laws apply. Dubai Police reported shops posting false reviews in October 2026.

Do it on your site: Add a plain-language privacy notice that names the law that applies to you (for example Tunisia's Organic Law 2004-63) and a contact address.

Sources: Baker McKenzie: Saudi Arabia's new Copyright Law <https://www.bakermckenzie.com/en/insight/publications/2026/07/saudi-arabia-new-copyright-law-modernises-ksa-ip-framework>; Latham & Watkins: Saudi Arabia's new copyright regulations, what AI companies should know <https://www.lw.com/en/insights/saudi-arabia-new-copyright-regulations-what-ai-companies-should-know>; Kennedys: Saudi PDPL enforcement decisions (title only, URL not verified); DLA Piper: Data protection laws, United Arab Emirates <https://www.dlapiperdataprotection.com/countries/uae-general/law.html>; Directinfo: ARP committee examines bills (Tunisia, data protection bill 095/2025) <https://directinfo.webmanagercenter.com/2026/05/20/tunisie-arp-la-commission-des-droits-et-des-libertes-examine-une-serie-de-projets-de-loi/>; Gulf News: Dubai Police on shops posting false reviews (title only, URL not verified)

### 13.5 Clicks and citations: the latest data (Evidence-based)

- Seer Interactive (2026 update, 53 brands, 5.47 million queries): the organic click rate on queries with an AI Overview recovered from 1.31% (Dec 2025) to 2.36% (Feb 2026), still below 3.82% for queries without one.
- Being cited inside the AI Overview brought about 120% more organic clicks per impression than not being cited. Seer says it cannot claim causation.
- Ahrefs (October 2026, 66.6 million ChatGPT answers): Reddit and Wikipedia citations fell sharply in US answers while primary sources such as PubMed and brand sites rose. Other datasets disagree, so measure your own appearance rate.

Do it on your site: Check whether your key page is cited in Google's AI Overview for your main question. If not, compare it with the cited pages.

Sources: Seer Interactive: AIO Impact on Google CTR, 2026 Update (53 brands, 5.47M queries) <https://www.seerinteractive.com/insights/aio-impact-on-google-ctr-2026-update>; Ahrefs: ChatGPT citation study, 66.6M responses (title only, URL not verified); Semrush: Most cited domains in AI <https://www.semrush.com/blog/most-cited-domains-ai/>

## Myth or fact? Test yourself

- Blocking GPTBot removes you from ChatGPT search. **Myth.** GPTBot is for training. ChatGPT search depends on OAI-SearchBot.
- Blocking Google-Extended removes you from AI Overviews. **Myth.** AI Overviews depend on Googlebot and snippet rules, not Google-Extended.
- Google Search ignores llms.txt. **Fact.** Google: such files neither harm nor help visibility in Google Search.
- FAQ schema still gives you dropdowns in Google. **Myth.** FAQ rich results stopped appearing on 7 May 2026.
- A tool can tell you your exact ranking position in ChatGPT. **Myth.** Answers change nearly every run; only appearance rates are meaningful.
- Copilot cites sources from Bing. **Fact.** Bing's AI Performance report covers Copilot citations.
- Keyword stuffing helps AI quote you. **Myth.** In the GEO study it performed worse than doing nothing.
- Public Instagram posts of professional accounts can appear in Google. **Fact.** Since 10 July 2025, for public professional accounts of adults.
- Buying reviews is legal if they look real. **Myth.** Banned by the FTC rule (2024) and the EU Omnibus Directive (2022).
- Search Console shows AI impressions since 2026. **Fact.** Generative AI reports launched in June 2026, worldwide since 31 August 2026.
- robots.txt is a technical lock that stops every bot. **Myth.** It is a voluntary standard; some user-triggered fetchers may not follow it.
- Your CDN can block AI bots even if robots.txt allows them. **Fact.** Google lists checking CDN and hosting settings as a best practice.
- Cloudflare's “Block AI bots” setting can also block Googlebot. **Fact.** Since 15 September 2026 it applies to mixed-use crawlers, including Googlebot, Bingbot and Applebot.
- In the US, robots.txt counts as an access control under the DMCA. **Myth.** OpenAI copyright litigation (S.D.N.Y., Dec 2025): disregarding robots.txt is not circumvention.
- Saudi Arabia allows copying lawfully published works to develop AI. **Fact.** Copyright Law 2026, Art. 26(4), in force since 12 August 2026.
