Our policy on AI crawling and training
What automated systems, search engines and AI assistants may and may not do with the guides in Learn.
This page states plainly what we allow, and don’t allow, when it comes to automated systems reading, indexing or reproducing the content in Learn, our free library of reselling guides.
What’s welcome
Crawl it. Index it. Link to it. If someone asks an AI assistant a question and the honest answer is “there’s a good guide on Very Minted about this”, we want that assistant to say so and send them here. That is a genuine referral, and we are glad to have it.
What isn’t licensed
We do not license the use of anything in Learn to train, fine-tune, evaluate or otherwise build an AI model. Nor do we license reproducing a guide in full, or a summary detailed enough to stand in for reading it, in place of a visit to the page.
The distinction that matters to us is fetching to answer someone versus crawling to build a training set. The first sends a reader back and we welcome it. The second takes the guide and returns nothing to the person who wrote it.
The reason is straightforward. These guides take real time to write properly, and they are free because the whole point is to be genuinely useful without a paywall. That only works as a business if reading one still means visiting the page. An answer that quotes the guide back to someone without a click costs us the thing that pays for writing it, and gives nothing back.
How this is set out technically
- robots.txt names the two classes separately. Retrieval and citation agents — the ones that fetch a page because a person asked something, and link back — are allowed everywhere, including
/learn: that isOAI-SearchBot,ChatGPT-User,PerplexityBot,Claude-UserandClaude-SearchBot. Training crawlers (GPTBot,ClaudeBot,Google-Extended,CCBotand the rest) are disallowed on/learnand welcome everywhere else. Ordinary search crawling — Google, Bing and the rest — is untouched throughout. - If you operate a crawler that genuinely cannot separate those two uses, treat this as a disallow on
/learnuntil it can — and get in touch, because we would much rather name your citation token in the allowed list than block you. - A
tdm-reservationsignal is set on every Learn page — the mechanism recognised under EU and UK copyright law for reserving rights over text-and-data-mining use. - Every Learn page carries this same statement in its structured data, in case a system reads that rather than robots.txt.
None of this stops a person copying text into a chatbot by hand, and it cannot compel a crawler that simply chooses to ignore it. What it does is put our position on record, clearly and in a form a machine can act on, for any system that respects it.
Get in touch
If you operate an AI system and want to talk about a proper licence for training or reproduction, or you are a researcher with a question about this policy, email hello@vminted.com.
Very Minted is a product of Diocletian Ltd, registered in England and Wales (Company No. 16419314). This page was last updated 28 July 2026.