About trainedOnMe
Why we built this
More and more people get their news from AI answers instead of from the publisher. Ad-supported publishers lose the search traffic that funds them, and subscription publishers see their exclusive reporting turn up in chatbots for free. Blocking AI crawlers protects only the publisher’s own site: the same stories still reach the models through other sites that carry them.
Publishers know their content is getting through, but they have no evidence. We built trainedOnMe to provide it: the answer, the facts in it that only you published, the pages it pointed to, and whether you were credited. That evidence helps price a licensing deal with an AI company, or show the harm in a legal claim. We work with publishers’ legal, IP and business development teams.
Who we are
We are Sahaj Saini and Alessandro Balzi. We met at Stanford University, where we found we shared the same interest in AI transparency and accountability.
trainedOnMe began as a research project. Over ten weeks we interviewed 83 people across news, music, AI labs and intellectual property law about how creative work is used to build AI, and who gets paid for it. The same gap kept coming up: the people who make original work had no way to see what AI models do with it.
So we built the tools to see it. First a test for whether a text was in a model’s training data, then a way to see when AI answers use a publisher’s reporting and whether they credit it. Today we run it with news publishers.
If you have questions or want to get in touch, use the contact form.