Gallery photos and AI training: robots.txt on Shopify

On 30 October 2026 Google announced that it plans to support a shared AI training control in robots.txt: one policy token, written once, that participating AI companies will read as a site's preference about whether its content may be used to train models. Microsoft and OpenAI said the same week that they would honour it. Support is promised for 2027, and until then the company-specific tokens, such as Google-Extended, keep working. Photographers, makers and brands who publish galleries of their own work on Shopify have been asking what this means for them. This guide goes through what a Shopify store can actually set, what it covers, and the one detail about where Snug Gallery images live that changes the answer.

What Google announced

The announcement builds on work at the IETF, where a group called AIPREF has been defining a vocabulary for expressing preferences about AI use of web content. Google says it will read a shared policy token in robots.txt and, where a site uses it to opt out of training and has not said anything specific about Google-Extended, treat the site as if it had disallowed Google-Extended. Google-Extended is the token Google has offered since 2023 for exactly this purpose; it controls whether content is used to train Gemini models and does not affect whether pages appear in Google Search.

Two things in the announcement matter for a store owner. First, it is a plan, with support promised in 2027, so there is nothing to write today beyond the existing company-specific tokens. Second, it is a preference, not a lock. Robots.txt tells well-behaved crawlers what you would like; it does not stop a crawler that ignores it, and Shopify's own help pages say the same about crawler rules on a store.

What a Shopify store can put in robots.txt

Every Shopify store serves a default robots.txt at the root of its primary domain, and Shopify's help page on editing robots.txt.liquid explains how to override it: add a robots.txt.liquid template to your theme in the code editor and write your rules there. The page lists what you can do, including allowing or disallowing URLs and blocking specific crawlers, and it carries a caution worth repeating: this is an unsupported customisation, Shopify Support will not help with it, and a mistake can cost you all of your search traffic.

Blocking an AI training crawler is a short addition. A rule for a named user agent, Google-Extended for Google, with a Disallow for the whole site, placed above or below the default rules, is the whole of it. When the shared token arrives it will be another line in the same file. The help page on agentic storefronts adds a point that is easy to miss: rules in robots.txt affect open-web discoverability only. If you have activated agentic sales channels, Shopify Catalog sends your product data to them regardless of what robots.txt says.

If you edit the template, test it. Open your store's /robots.txt in a browser after saving and read it top to bottom. A stray rule that disallows everything for every user agent looks almost identical to one that disallows everything for one named crawler, and only one of them is what you meant.

Where your gallery images actually live

Here is the detail. Snug Gallery stores every photo and video you upload in your store's Files, and the storefront serves them from Shopify's content delivery network at cdn.shopify.com, not from your domain. The guide on who owns your gallery images explains why that is the right design for ownership and speed. It also means your robots.txt does not govern the image files. A robots.txt file applies to the host it is served from; yours is at yourstore.com/robots.txt and says nothing about cdn.shopify.com, which has its own, maintained by Shopify, that you cannot edit.

What your robots.txt does govern is the pages. The gallery page on your domain is where the image URLs appear, together with the alt text, the caption titles and the caption text that Snug Gallery writes into the HTML. A crawler that honours a disallow for your pages never reads that HTML and never learns the image URLs from it. A crawler that already has the URLs from somewhere else, or that reaches cdn.shopify.com by other means, is outside your file's reach.

The same applies to your product images, your theme's images and anything else in Files. This is not a gallery-specific limitation; it is how Shopify hosts media for every store. If a precise answer about cdn.shopify.com and AI crawlers matters to you, Shopify is the party to ask.

What that means in practice

For most stores the honest summary is: a robots.txt rule expresses your preference to the companies that have said they will honour it, covers the pages on your domain, and reaches the images only indirectly, by not handing their URLs to a crawler that respects the rule. That is still worth doing if the preference matters to you. It is a few lines, it is reversible, and the companies involved have publicly committed to reading it.

It is not a protection in the sense the right-click protection guide is careful to explain for human visitors either: anything a browser can display can be copied, and anything a browser can fetch a crawler can fetch. Treat robots.txt as a label on your work, not a lock on the door.

Think about what else publishes your images. The Pinterest save button, if you have it on, exists to send images to Pinterest. Google Images indexes gallery photos from pages it is allowed to crawl, and the image search guide shows how to make that work for you. A gallery that is good at being found is, by the same mechanisms, visible to any crawler that is allowed to see the page. Decide which you want; you can have a page that is open to search and still asks AI trainers to stay out, because Google-Extended and Googlebot are separate tokens.

The trade-off Google states plainly

The announcement notes that websites that specifically allow Google-Extended are eligible to appear in responses, with clicks, through grounding in Gemini apps. Google is telling you the price of opting out: your pages may not be cited when someone asks an assistant where to buy the kind of thing you make. For a store that sells, rather than a photographer protecting a portfolio, that is a real cost, and it argues for being specific rather than blanket. You can disallow training on a portfolio page and leave product and collection pages open, because robots.txt rules are per path.

Snug Gallery helps with that split if you keep different kinds of images on different pages. A portfolio or lookbook page with your original photography can sit under a path you disallow for training crawlers, while a product page gallery of the things you sell stays open. The gallery block does not know or care which rules apply to the page it is on.

A short checklist

Decide whether you want to express a training preference at all, and for which pages. Add a robots.txt.liquid template only if the answer is yes, and keep the default rules intact above your additions. Add a Google-Extended disallow for the paths you chose; add the shared token when Google and the others publish the exact syntax in 2027. Open /robots.txt and read it. Check Google Search Console a week later to confirm Googlebot is still crawling. And remember that the images themselves are on cdn.shopify.com, where your file does not apply, so the rule protects the page, the alt text and the captions directly, and the images only through them.

Frequently asked questions

Can I block AI crawlers from my Shopify gallery images with robots.txt?

You can ask them to stay off your pages. Add a robots.txt.liquid template to your theme with a Disallow rule for the crawler's user agent, such as Google-Extended, and crawlers that honour robots.txt will not read the gallery page or the image URLs in it. The image files themselves are served from cdn.shopify.com, which has its own robots.txt that you cannot edit, so the rule reaches them only through your pages.

Does blocking Google-Extended remove my gallery from Google Images or Google Search?

No. Google-Extended is a separate token from Googlebot and controls use for training Gemini models only. Google states it does not affect inclusion or ranking in Search. A page can stay fully indexed, including its images, while asking Google not to use it for training. Google does note that sites which allow Google-Extended are eligible to be cited in Gemini app responses.

What is the shared AI training control Google announced in October 2026?

A single policy token in robots.txt, based on the IETF AIPREF work, that participating AI providers will read as a site's training preference. Google, Microsoft and OpenAI have said they will honour it, with Google's support planned for 2027. If a site opts out with the token and does not mention Google-Extended, Google will treat it as a Google-Extended disallow. The exact syntax is not final yet.

Does Snug Gallery add any crawler rules of its own?

No. Snug Gallery writes the gallery into your theme as HTML with images on cdn.shopify.com and does not touch robots.txt, meta robots tags or any crawler settings. Whatever you set in robots.txt.liquid applies to gallery pages exactly as it applies to the rest of your store.