--- title: The courts priced provenance | Xferdata description: A $1.5bn settlement turned on how training data was obtained, not what was done with it. That is why a buyer interrogates your right to collect and right to share. image: https://www.xferdata.com/hubfs/xferdata-blog-p8-provenance-priced.png --- [XFERDATA](https://www.xferdata.com/) [For owners](https://www.xferdata.com/owners) [For buyers](https://www.xferdata.com/buyers) [Blog](https://www.xferdata.com/blog) [Contact](https://www.xferdata.com/contact) [Get a valuation](https://app.prod.xferdata.com/auth/login?src=nav-valuation) ![A settlement that turned on how the books were obtained, not what was done with them](https://www.xferdata.com/hubfs/xferdata-blog-p8-provenance-priced.png) # The courts priced provenance Xferdata  ·  Oct 21, 2026, 9:00:00 AM In September 2025 Anthropic settled with a group of authors for $1.5 billion, one of the largest copyright settlements in US history. The number is not the interesting part. What the case turned on is. Three months earlier, Judge William Alsup had found that training a language model on books was a plausible fair use, because what the model does with the text is transformative. That part went in the AI company's favour. The case proceeded anyway, on a separate question: where the books came from. Lawfully bought books and pirated books were treated differently, and it was the acquisition that carried the exposure. Read that as a market signal rather than a legal one. A court effectively said that what you do with data and how you got hold of it are two different questions, and the second one can be worth more than the first. ## Why a buyer now asks you awkward questions If you take a dataset to market, expect to be asked where every field came from, whether it was captured or derived, and what your terms of service, privacy notices and customer contracts actually permit. This is not diligence theatre. A buyer who licenses data you did not have the right to license inherits a problem that no indemnity fully solves, because the counterparty who objects is not you, it is the customer or supplier whose data it is. After a $1.5bn settlement, every serious buyer has a lawyer who has read the ruling. The practical effect is that provenance moved from a footnote in the pack to something that gets priced. Clean provenance is not a bonus. Murky provenance is a discount, and often a refusal. ## The two rights people conflate There is a right to collect and there is a right to share, and having the first does not give you the second. A company can lawfully collect transaction records for the purpose of running its service and have no right at all to license those records to a third party, because nothing in the terms the customer agreed to contemplates it. This is the single most common reason a dataset that looked valuable turns out not to be sellable. The fix is unglamorous and it is mostly drafting. Terms of service that anticipate aggregated and anonymized use. Privacy notices that describe it plainly. Supplier contracts that do not quietly assign the rights elsewhere. None of this is expensive if you do it before a buyer asks. All of it is slow if you start after. ## What this is worth to you The uncomfortable framing is that the AI industry spent a decade treating acquisition as somebody else's problem, and the bill arrived. The comfortable framing is that a market which now pays attention to how data was obtained is a market that rewards companies who can prove it. If your rights position is clean, you are competing against a lot of datasets whose owners have never checked. That is worth more than it was two years ago. ## This is the market we run Xferdata values a company's data, builds the auction pack, and runs the auction to buyers globally. **If you own the data**, run a valuation and find out what yours is worth. **If you buy data**, sign up and tell us what you are after. You will see packs as they come to market. [Sign up to Xferdata](https://app.prod.xferdata.com/auth/login?utm_source=blog&utm_medium=organic&utm_campaign=data_auctions_2026q4&utm_content=courts_priced_provenance) © 2026 Xferdata, Inc. All rights reserved. [For owners](https://www.xferdata.com/owners) [For buyers](https://www.xferdata.com/buyers) [Blog](https://www.xferdata.com/blog) [Contact](https://www.xferdata.com/contact) [Privacy](https://www.xferdata.com/privacy) [LinkedIn](https://www.linkedin.com/company/xferdata) × Loading calendar… ```json { "@context" : "https://schema.org", "@type" : "BlogPosting", "author" : { "@type" : "Person", "name" : "Xferdata", "url" : "https://www.xferdata.com/blog/author/xferdata" }, "dateModified" : "2026-10-21T13:00:00.260Z", "datePublished" : "2026-10-21T13:00:00.000Z", "headline" : "The courts priced provenance", "image" : [ "https://www.xferdata.com/hubfs/xferdata-blog-p8-provenance-priced.png" ], "mainEntityOfPage" : { "@id" : "https://www.xferdata.com/blog/the-courts-priced-provenance", "@type" : "WebPage" }, "publisher" : { "@type" : "Organization", "logo" : { "@type" : "ImageObject", "url" : "https://www.xferdata.com/hubfs/Brand%20Logos/xferdata-logo-navy-blue.svg" }, "name" : "Xferdata" } } ``` ```json { "@context" : "https://schema.org", "@type" : "Organization", "description" : "Xferdata values enterprise data assets and runs competitive auctions in which institutional buyers license them.", "email" : "info@xferdata.com", "logo" : "https://47529209.fs1.hubspotusercontent-na1.net/hubfs/47529209/xferdata-favicon.png", "name" : "Xferdata", "sameAs" : [ "https://www.linkedin.com/company/xferdata" ], "url" : "https://www.xferdata.com" } ```