AI and training data · Brief

Publishers agree a common language for licensing content to AI

Really Simple Licensing 1.0 lets a website state, in a form machines can read, how AI may use its work and on what terms.

10 December 2025 · 2 min read

A row of old brass keys hanging on hooks in a sunlit wooden key cabinet

Picture made for ABURA Bureau.

On 10 December 2025 the RSL Technical Steering Committee published the official specification for Really Simple Licensing 1.0, a standard for telling AI systems how a website's content may be used. It was developed with publishers and platforms including Yahoo, Ziff Davis and O'Reilly Media.

Robots.txt, the file most websites use to talk to crawlers, can only say yes or no. RSL adds machine-readable licensing terms on top of it: which uses are allowed, and whether payment or credit is required. The first full version lets a site stay in search results while opting out of AI search applications, and lets owners of free content ask AI systems that benefit from it for a contribution.

A standard like this works only when AI developers honour it, or when the companies that deliver websites enforce it. Its value for owners is the vocabulary. Training a model, answering questions and building a search index are different uses, and they can carry different prices. A licence that names each use is easier to price, easier to check and easier to renew than one blanket permission.

Looking for data the internet does not have?

Tell us what you are training and we will show you what owners are offering.

Start on abura.io

Get the weekly issue

One email a week once the issue starts, only the desks you pick. Leave whenever you like.

Your desks

We use your email only to send the issue. Your rights.

More from AI and training data

A vintage reel of magnetic computer tape on a steel desk, the tape unspooling into ribbons of luminous code that rise and weave into a translucent glass neural latticeAI and training data
AI and training data · Explainer

Why old code is the hottest data in AI right now

AI teams want code that people wrote before AI started writing it. Many companies have years of it sitting in their repositories.

19 October 2026 · 2 min read
An ancient ochre handprint on a cave wall, the pigment lifting off the rock as thousands of fine glowing threads that flow into a floating crystal coreAI and training data
AI and training data · Explainer

Why AI still needs data made by people

AI teams are running short of good human-made data, and the open web is closing its doors. Licensed company data is one of the few new sources left.

19 October 2026 · 2 min read
A tall wall of wooden card-catalogue drawers in an old archive, one drawer pulled openAI and training data
AI and training data · Brief

Europe sets the form AI developers must use to describe their training data

The European Commission publishes the template general-purpose AI models must fill in under the AI Act, with its own place for licensed data.

24 July 2025 · 2 min read