--- title: The Problem of Provenance: Nobody Knows What’s In the... description: Data provenance in AI training is a systemic problem: most large models are trained on composite datasets with no reliable record of origin, collection date,... canonical: https://daily.dev/posts/the-problem-of-provenance-nobody-knows-what-s-in-the-training-data-p0me2twe9 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: The Problem of Provenance: Nobody Knows What’s In the Training Data | daily.dev og:description: Data provenance in AI training is a systemic problem: most large models are trained on composite datasets with no reliable record of origin, collection date,... og:url: https://daily.dev/posts/the-problem-of-provenance-nobody-knows-what-s-in-the-training-data-p0me2twe9 og:image: https://api.daily.dev/og/posts/p0mE2TWE9.png og:image:alt: The Problem of Provenance: Nobody Knows What’s In the Training Data og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # The Problem of Provenance: Nobody Knows What’s In the Training Data **[Medium](https://daily.dev/sources/medium_js)** · 4 min read · 0 upvotes · 0 comments ## Summary Data provenance in AI training is a systemic problem: most large models are trained on composite datasets with no reliable record of origin, collection date, or modification history. Research from the Data Provenance Initiative found over 70% of popular dataset licenses on GitHub and Hugging Face are unspecified, and 66% of labeled ones are miscategorized as more permissive than intended. This creates legal, quality, and safety risks — biased or toxic outputs can't be traced, contamination hides more easily, and companies unknowingly inherit licensing liability. A proposed fix borrows from software engineering: an ML-BOM (Machine Learning Bill of Materials), a standardized, machine-readable record of what data went into a model and under what terms. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What percentage of AI training dataset licenses are unspecified or mislabeled? More than 70% of licenses for popular datasets hosted on platforms like GitHub and Hugging Face are unspecified. Among datasets that were labeled, a re-annotation found 66% were miscategorized, typically listed as more permissive than the original creator intended, based on an analysis of over 1,800 datasets by the Data Provenance Initiative. _Anyone vetting datasets for fine-tuning can follow provenance and licensing research on daily.dev._ ### What is an ML-BOM and how does it relate to a software bill of materials? An ML-BOM is a proposed documented, machine-readable record of what data went into a machine learning model, where it came from, and under what license terms. It mirrors the Software Bill of Materials (SBOM) concept, which lists the components, licenses, and dependencies in a piece of code, applying the same idea of traceable inputs to AI training data. _Engineers tracking emerging AI supply-chain standards like ML-BOM can follow the discussion on daily.dev._ ## Similar posts on daily.dev - [Why AI Is Training on Its Own Garbage \(and How to Fix It\)](https://daily.dev/posts/why-ai-is-training-on-its-own-garbage-and-how-to-fix-it--mkbzwgndw) · Towards Data Science · 0 upvotes · 0 comments - [Introducing Model Provenance Kit: Know Where Your AI Models Come From](https://daily.dev/posts/introducing-model-provenance-kit-know-where-your-ai-models-come-from-vc0ij2exo) · Cisco · 1 upvotes · 0 comments - [The next AI breakthrough won’t come from bigger models, but from better data](https://daily.dev/posts/the-next-ai-breakthrough-won-t-come-from-bigger-models-but-from-better-data-8ruccmjsi) · InfoWorld · 1 upvotes · 1 comments --- Tags: [#machine-learning](https://daily.dev/tags/machine-learning), [#ai-safety](https://daily.dev/tags/ai-safety) [View this post on daily.dev](https://daily.dev/posts/the-problem-of-provenance-nobody-knows-what-s-in-the-training-data-p0me2twe9) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"The Problem of Provenance: Nobody Knows What’s In the Training Data","url":"https://daily.dev/posts/the-problem-of-provenance-nobody-knows-what-s-in-the-training-data-p0me2twe9","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/the-problem-of-provenance-nobody-knows-what-s-in-the-training-data-p0me2twe9"},"datePublished":"2026-07-22T17:09:31.650Z","dateModified":"2026-09-13T19:03:50.688Z","description":"Data provenance in AI training is a systemic problem: most large models are trained on composite datasets with no reliable record of origin, collection date,...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/13d39c7802ada70f4762d3fdaced9ada?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/13d39c7802ada70f4762d3fdaced9ada?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Medium","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Medium","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/medium","url":"https://daily.dev/sources/medium_js"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/the-problem-of-provenance-nobody-knows-what-s-in-the-training-data-p0me2twe9","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"machine-learning,ai-safety","timeRequired":"PT4M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Medium","item":"https://daily.dev/sources/medium_js"},{"@type":"ListItem","position":3,"name":"The Problem of Provenance: Nobody Knows What’s In the Training Data"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/the-problem-of-provenance-nobody-knows-what-s-in-the-training-data-p0me2twe9#faq","mainEntity":[{"@type":"Question","name":"What percentage of AI training dataset licenses are unspecified or mislabeled?","acceptedAnswer":{"@type":"Answer","text":"More than 70% of licenses for popular datasets hosted on platforms like GitHub and Hugging Face are unspecified. Among datasets that were labeled, a re-annotation found 66% were miscategorized, typically listed as more permissive than the original creator intended, based on an analysis of over 1,800 datasets by the Data Provenance Initiative. Anyone vetting datasets for fine-tuning can follow provenance and licensing research on daily.dev."}},{"@type":"Question","name":"What is an ML-BOM and how does it relate to a software bill of materials?","acceptedAnswer":{"@type":"Answer","text":"An ML-BOM is a proposed documented, machine-readable record of what data went into a machine learning model, where it came from, and under what license terms. It mirrors the Software Bill of Materials (SBOM) concept, which lists the components, licenses, and dependencies in a piece of code, applying the same idea of traceable inputs to AI training data. Engineers tracking emerging AI supply-chain standards like ML-BOM can follow the discussion on daily.dev."}}]} ```