Direct answer: yes. Google does not treat AI or machine translation as spam in itself, and no UK rule we found bans machine-translated travel pages. But a translated page is held to the same standard as the original. Google’s spam policy lists “automated transformations like synonymizing, translating, or other obfuscation techniques” among examples of scaled content abuse “where little value is provided to users” (Google spam policies). Google’s Gary Illyes, asked whether auto-translating a site hurts rankings, said: “if the auto translation is of low quality, maybe”, and that a human native speaker should review “(and perhaps fixes)” the translations (Google SEO office hours, June 2024). And UK consumer law does not relax because a page is in German or Spanish: pre-contract information must still be given “in a clear and comprehensible manner” (Consumer Contracts Regulations 2013, regulation 13), and a mistranslated price, inclusion or cancellation term is still “false or misleading information relating to a product” (DMCC Act 2024, section 226).
Kaize is a UK AI product studio for travel and hospitality. This article is based on Google Search Central documentation, Google Hotel Center developer documentation, Google Cloud Translation and DeepL product documentation, UK legislation and the EU AI Act text, all read on 16 October 2026. It is not legal advice or a case study; the controls below are our recommendations.
Why this question is live now
Translation used to be a cost line that kept most UK hotels, villa companies, tour operators and DMCs to English, plus perhaps one or two languages done by an agency. Large language models and translation APIs have cut much of that cost: a CMS plugin or booking-engine setting can now produce every room page, tour itinerary and FAQ in a dozen languages overnight. The question has moved from “can we afford to translate?” to “what happens to search visibility, guest expectations and our legal obligations when we do it at scale?”
The rules that answer it are not new, but Google keeps them current. Its generative AI content guidance shows a last update of 1 October 2026, its page on localized versions and hreflang 21 September 2026, its helpful-content guidance 5 October 2026 and its spam policies 28 August 2026 (Google generative AI content guidance; localized versions; helpful content; spam policies). Read together, they give a clear answer for travel businesses — with a consumer-law layer on top that Google does not cover.
What Google actually says about AI-translated pages
There is no machine-translation penalty and no special label. Asked how to be transparent about AI translations, Google’s John Mueller said “there’s no special markup that you can add to your pages to label them as automatic translations”, and: “If the pages are well-translated, if it uses the right wording for your audience, in short, if you think they’re good for your users, then making them indexable is fine. If you’re not happy with them, you could just include the noindex robots meta tag on them” (Google SEO office hours, June 2024). He added that “a good localization is much more than just a translation of words and sentences.”
The risk is scale without value. Google defines scaled content abuse as “when many pages are generated for the primary purpose of manipulating search rankings and not helping users”, “no matter how it’s created”, and says that if you host such content, “exclude it from Search” (Google spam policies). A hotel that machine-translates its genuinely useful room and policy pages for guests who read that language is not doing what the policy describes. A tour operator that spins one itinerary into thirty thin language-and-city variants to catch long-tail searches is much closer to it.
Review is expected, including the invisible parts. Google says “it is critical to manually factcheck and review all AI-generated content for accuracy and trustworthiness before publishing”, and that this applies to title elements, meta descriptions, structured data and image alt text, all of which can appear in results (Google generative AI content guidance). Translated pages are where those fields most often get skipped: the body is reviewed, the meta description and the structured data stay as raw machine output.
Google’s helpful-content guidance also asks site owners to think about “Who, How, and Why”, including: “Is the use of automation, including AI-generation, self-evident to visitors through disclosures or in other ways?” (Google helpful content guidance). That is a question to answer for your audience, not a mandatory label.
First decision: do you need your own translated page at all?
Google may already be translating for you. Its translated results feature means that “sometimes Google may translate the title link and snippet of a search result for results that aren’t in the language of the search query”; a user who clicks is shown “a page that’s been machine translated”, and “Google doesn’t host any translated pages” (Google translated results). Google lists the languages it may translate into, including French, German, Spanish, Portuguese, Arabic, Korean and Turkish. On the page we read, Italian, Dutch, Japanese and Chinese were not on that list. Publishers can opt out of all Google translation features with the notranslate rule, and clicks and impressions from translated results can be monitored with the Search Appearance filter in Search Console’s Performance report.
That gives a practical way to decide where your own translated pages are worth the effort:
- Build a dedicated language version where a market matters commercially and you can do better than an on-the-fly translation: local currency and payment, correct room and board terminology, local-language customer service, and legal and booking information you have actually checked.
- Rely on English plus Google’s translated results (or browser translation) for languages where you cannot yet review or support guests — and accept that you do not control the wording guests see.
- Do not publish languages you cannot answer guests in. A German booking page that leads to an English-only phone line or confirmation email creates exactly the expectation gap consumer law cares about.
- Record the decision per market, with an owner, so new languages are added deliberately rather than because a plugin made it possible.
The technical setup Google expects
Separate URLs per language. Google “recommends using different URLs for each language version of a page rather than using cookies or browser settings to adjust the content language on the page”, and warns that if you switch content dynamically, it “might not find and crawl all your variations” because “the Googlebot crawler usually originates from the USA” and “sends HTTP requests without setting Accept-Language in the request header” (Google: managing multilingual sites). Google’s page on locale-adaptive pages repeats that a separate-URL configuration with hreflang annotations is recommended (locale-adaptive pages). Some CMS translation plugins and booking engines switch language by cookie or browser setting on a single URL; check how yours works before you scale it.
hreflang, done completely. hreflang tells Google which pages are language or regional variants of each other. “Each language version must list itself as well as all other language versions”, alternate URLs must be fully qualified, and “if two pages don’t both point to each other, the tags will be ignored”; Google also recommends an x-default fallback for unmatched languages (Google: localized versions). It is not a language detector: Google says it doesn’t use hreflang or the HTML lang attribute to detect a page’s language, but uses algorithms.
One language per page, and no forced redirects. Google “uses the visible content of your page to determine its language”, recommends a single language for content and navigation on each page, warns that translating only the boilerplate while the main content stays in another language “can create a bad user experience”, and says to avoid “automatically redirecting users from one language version of a site to a different language version” — offer language links instead (Google: managing multilingual sites). For travel sites the common failure is a translated template wrapped around untranslated room descriptions, guest reviews or supplier copy.
For hotels on Google, align Hotel Center. Hotel price feeds use a points-of-sale file to define landing pages, and its Match element can filter on “the user and hotel’s language” using a two-letter language code (Google Hotel Center points of sale syntax). If you add a French booking page, decide deliberately whether French-language users should land on it — and check the rates and conditions on that page match the feed. Our earlier piece on Hotel Center price accuracy covers the price side.
What must never be machine-translated unreviewed
Google’s rules decide whether a translated page is worth indexing. UK consumer law decides whether what it says to a guest is lawful. Before a distance contract, the trader must provide the required information “in a clear and comprehensible manner” (CCR 2013, regulation 13). Written consumer contract terms and notices must be “transparent”, meaning “expressed in plain and intelligible language” and legible (Consumer Rights Act 2015, section 68). Package information must be provided “in a clear, comprehensible and prominent manner”, and the standard information under Schedule 2 must use “the form and wording set out in that Schedule” (Package Travel Regulations 2018, regulation 5). And under the DMCC Act, misleading information includes information “which, although true, is presented in a misleading way” (DMCC Act 2024, section 226).
None of those provisions mention translation. They apply to whatever the guest reads. That is why we recommend splitting content into tiers before any translation job runs:
- Tier 1 — inspiration and destination content (area guides, blog posts, experience descriptions without prices): AI translation with a terminology glossary and a native-speaker review of a sample before indexing.
- Tier 2 — product facts (room types and sizes, bed configurations, board basis, what a tour includes, accessibility features, allergen and dietary information, transfer times, age limits): AI translation is a draft; every page is reviewed by a native speaker against the English source before it is indexable.
- Tier 3 — price, legal and booking conditions (total price and mandatory fees, deposits, cancellation and refund terms, package standard information, insolvency protection wording, terms and conditions, privacy notice): human-approved translations stored as locked strings, versioned, and never regenerated on page build. If wording is prescribed, the approved translation is the only one that ships.
- Out of scope for automatic translation: anything personalised and transactional that is generated per booking, unless that pipeline has its own controls.
If you also sell into the EU, note the EU AI Act’s rule on AI-generated text: deployers must disclose text that “is published with the purpose of informing the public on matters of public interest”, unless it “has undergone a process of human review or editorial control” and someone holds editorial responsibility (EU AI Act, Article 50(4)). Ordinary hotel and tour marketing pages are unlikely to be that kind of text, and a reviewed translation workflow is consistent with the exception either way. Our article on AI disclosure in guest chat under Article 50 covers the chatbot side.
Glossaries: the control most translation projects skip
Most translation errors that matter in travel are terminology errors, not grammar. Translation APIs support glossaries for exactly this: Google describes a glossary as “a custom dictionary the Cloud Translation API uses to consistently translate the customer’s domain-specific terminology” (Google Cloud Translation glossaries), and DeepL says its glossaries “allow you to fine-tune how words and phrases should be translated from one language to another” (DeepL glossaries guide). An LLM-based workflow needs the same thing as a term list passed with every request and checked afterwards.
A travel glossary should at least fix:
- Property, brand, room-type and package names — usually not translated at all.
- Board basis and inclusions: room only, bed and breakfast, half board, full board, all inclusive — and what each actually includes at your property.
- Price and fee terms: non-refundable, deposit, balance due, resort or service fee, tourist tax or visitor levy, per person versus per room.
- Protection and regulatory terms: ATOL, ABTA, package, linked travel arrangement — usually kept in English with an approved explanation.
- Place names and transfer points where the local and English names differ.
- Accessibility terms (step-free, accessible bathroom, roll-in shower) that guests rely on and models often soften.
A worked example (hypothetical)
A UK boutique hotel group wants German and French versions of its site before the next booking season. A CMS plugin offers one-click translation of all 400 pages. Following the steps above, the group decides that German justifies its own pages (it has German-speaking reservations staff) and French will, for now, rely on English plus Google’s translated results. It moves from cookie-based language switching to /de/ URLs with reciprocal hreflang and an x-default, and stops redirecting visitors by browser language. It builds a 60-term German glossary, translates Tier 1 pages with a sample review, has every room page reviewed by its German-speaking staff, and commissions an approved translation of its booking conditions and cancellation policy that is stored as fixed text. Translated room pages stay noindex until reviewed. During review, a staff member catches the machine translation rendering “breakfast available” as breakfast included on three room types — the kind of error that becomes a complaint, a chargeback or a misleading-information problem if it ships.
A pre-index gate for every translated page
- The page has its own URL, a single language in visible content and navigation, and reciprocal hreflang including itself, with x-default set.
- No automatic redirect by browser language or IP; a visible language switcher links to the other versions.
- Terminology glossary applied; brand, room and package names unchanged.
- Tier 2 facts checked against the English source by a native speaker; reviewer and date recorded.
- Tier 3 text is the approved, versioned translation — not machine output — and matches the current English version.
- Title, meta description, structured data and image alt text reviewed in the target language.
- Prices, fees and conditions match the booking engine, feeds and confirmation emails for that language and currency.
- Guest support exists in that language, or the page says clearly which language support is offered in.
- If any check fails, the page stays noindex until fixed.
How to measure whether it worked
Measure each language as a market, not as page volume. In Search Console, compare impressions and clicks by page and country for the new language folder against its own baseline, and use the Search Appearance filter to see how much traffic already arrives through Google’s translated results for that language. Track bookings and enquiries by site language, and watch complaints and guest-service tickets for terminology errors. Indexed page count is not a success metric; it is exactly the number scaled content abuse policies are written about.
What this article is not
- Not legal advice, and not a review or ranking of any translation tool, plugin or provider.
- Not a claim that machine translation harms rankings — Google says quality and value decide, not the method.
- Not a case study; the example above is hypothetical and Kaize reports no client results here.
Questions
Do we have to label AI-translated pages?
Not for Google: there is no markup for it (Google SEO office hours, June 2024). We found no UK rule requiring a “machine translated” label on a travel website; the UK duties are about the accuracy and clarity of what the page says. Google’s helpful-content guidance does suggest considering whether your use of automation is evident to visitors (Google helpful content guidance).
Should unreviewed machine translations be noindexed?
That is what Google suggests if you are not happy with their quality: include the noindex robots meta tag (Google SEO office hours, June 2024). In our view the default for Tier 2 and Tier 3 content should be noindex until reviewed.
Can we serve every language on one URL and switch by browser setting?
Google recommends separate URLs per language, because Googlebot usually crawls from the USA without an Accept-Language header and may not see your variations (Google: managing multilingual sites).
Does this affect how AI assistants describe us in other languages?
We have no measurement showing how ChatGPT, Gemini, Claude or Perplexity use translated hotel pages, so we make no claim. What we can say is that clear, consistent, crawlable language versions remove one reason for an assistant to guess — and that blocking or confusing crawlers works against you; see our guide to AI crawlers and robots.txt.
Kaize point of view
Translate the product, not just the words. Machine translation is now good enough to be the first draft of almost every travel page, and refusing to use it leaves international guests with nothing or with Google’s on-the-fly translation. But a language version is a market launch: it sets expectations about rooms, inclusions, prices and rights that someone has to stand behind. Our view is that hotels, tour operators and DMCs should translate deliberately — fewer languages done properly rather than every language a plugin offers — with separate URLs and hreflang, a terminology glossary, native review of product facts, and locked, approved translations for anything that touches price, cancellation or protection. Do that and AI makes international distribution cheaper; skip it and you have published thirty versions of the same mistake.
Related reading: Google AI Mode hotel booking and UCP for lodging, AI crawlers, robots.txt and your CDN, AI-edited hotel and villa photos, total price in AI travel quotes under DMCC, CCR 2013 pre-contract information before AI completes a sale and unfair terms in AI-written booking notices.
If you run a hotel group, tour operator or DMC and want help planning multilingual content, a translation pipeline with glossaries and review gates, or language versions that stay consistent with your booking engine, start with an AI opportunity review or discuss a project with Kaize. For how content and bookings move between your systems, see connecting travel booking systems and CRM.