Is my writing used to train AI?

What exactly do the legal pages say?

The commitment in the Content and Intellectual Property Policy reads, in full: "In yazar.io's current architecture, no component processes your user content or sends it into an AI/large-language-model training pipeline. Your books, characters, or notes are never part of an AI training dataset." The same policy says the copyright in everything you create is yours, and that yazar.io never acquires the right to publish or commercialize your work or to create derivative works from it.

Under the Terms of Use, you don't transfer a broad, perpetual or commercial license to yazar.io. What you grant is a non-exclusive, limited technical license that lasts only while your account is active and covers only storing your content and showing it to you, backing it up, exporting it when you ask, and publicly displaying what you choose to publish on your Author Card or a Shareable Wiki. AI training isn't on that list.

That commitment is about what yazar.io itself does: it never hands your content to AI training and never shares what you keep private. Pages you publish on your wiki are open to other companies' crawlers like any page on the web; more on that below.

What happens to your writing inside yazar.io?

yazar.io has no AI features, and no model processes what you write. Your content is stored on Supabase's infrastructure, and the Privacy Policy says every service provider except Paddle processes data on yazar.io's behalf and under its instructions. No creative content, such as a character's name, a book's title or the text of a note, goes to the analytics tools (PostHog and Google Analytics). It's also deliberately kept out of Sentry, the error tracking service, and it never reaches Paddle, the payment provider.

The descriptions of your characters, places and other records, your notes and your custom field values are stored encrypted with AES-256-GCM, with the keys kept apart from the database and protected by Google Cloud's key management service (KMS). This isn't end-to-end encryption: when an authorized request comes in, for example when you sign in and open a record, the server decrypts the content. The encryption protects your content if the database, or a backup of it, is compromised on its own.

What changes when you publish your wiki?

Your books are visible only to you by default, and yazar.io never shares what you keep private. Once you publish a book's Shareable Wiki, the pages you choose become public like any other page on the web: anyone can find and read a Public wiki, and anyone with the link can open an Unlisted one. yazar.io doesn't block any crawler from published wiki pages. Search engines can index Public pages, and crawlers that collect AI training data can reach them, along with any Unlisted page whose link they come across. The same goes for an Author Card you publish.

The Terms of Use list search engines and AI-powered search and answer services crawling public pages in line with robots.txt as a permitted use, and they forbid using automated tools to systematically collect other users' wiki content or turn it into a dataset. That rule doesn't technically stop a crawler, though, and copies taken from an open page are outside yazar.io's control.

If you want something to stay with you alone, don't publish it: keep your wiki's visibility on Private, the default, or hide that section or record from the wiki. An Unlisted wiki stays out of the sitemap and asks search engines not to index it, but that's a request, and anyone with the link can open the pages. On Yazar Pro you can also add a password to an Unlisted wiki; visitors are asked for it before the pages open. Your Notes, Quotes and Boards never appear in a wiki.

What should you check in other tools?

For the app you write your manuscript in, the chatbot you brainstorm with or the site where you post chapters, search the privacy policy and terms for words like "train", "model" and "improve our services", and check whether the license you grant stops at providing the service. Then look at the settings. On a personal ChatGPT plan, for example, OpenAI's help center says turning off "Improve the model for everyone" under Settings > Data controls keeps new conversations out of model training, although if you rate a response with a thumbs up or down, OpenAI may still use that whole conversation to improve its models.

If you post stories on your own website, you can add rules for GPTBot and Google-Extended to your robots.txt file. OpenAI says disallowing GPTBot signals that a site's content shouldn't be used to train generative AI foundation models; Google-Extended controls whether content Google crawls may be used to train future Gemini models, and blocking it doesn't affect your site's inclusion or ranking in Google Search. Either way, robots.txt is a request: it won't stop a crawler that ignores it.