RSL: Machine-Readable Content Licensing for AI
RSL (Really Simple Licensing) attaches five machine-readable AI license models: free, attribution, subscription, pay-per-crawl, pay-per-inference.
RSL licenses content to AI machine-readably
RSL (Really Simple Licensing) is a machine-readable content-licensing standard that lets a publisher attach explicit terms to content for AI use, so the license travels with the content as data an agent can parse rather than as prose buried in a terms-of-service page. An RSL license is publisher-controlled and referenced from robots.txt (and discoverable per-resource), which lets it be enforced at the edge — for example by a CDN such as Cloudflare that reads the declared terms and gates access accordingly. RSL is a Layer-6 licensing protocol; its launch date, the exact set and names of the license models, and any adopter figure must be verified against the primary rslstandard.org spec at build — none of those are asserted from memory here.
RSL spans five license models from free to pay-per-inference
RSL is built around five business models, ordered roughly from most-open to most-metered:
- Free — use the content with no fee and no required credit.
- Attribution — use it, but credit and link the source.
- Subscription — flat-fee access for a licensee.
- Pay-per-crawl — charge per fetch of the content.
- Pay-per-inference — charge each time the content is used to generate an answer.
The model set per the project brief and the rslstandard.org spec is the authority for the names and definitions; confirm them at build.
Pay-per-inference prices the use, not just the fetch
The two usage-priced models look similar but bill at different moments. Pay-per-crawl prices the fetch — the publisher is paid when the content is retrieved. Pay-per-inference prices the use in an answer — the publisher is paid each time the content actually informs a generated response. Pay-per-inference therefore aligns the publisher's payment with the value an AI extracts from the content rather than with the act of crawling, which is attractive for high-value reference material that may be fetched once but cited many times. Because RSL is machine-readable and edge-enforceable, either model can be declared and collected without the publisher running its own metering or billing stack.
Related: RSL licenses content to AI across five models · RSL defined in the Lexicon · the pay-per-crawl mechanism explained in full · every mechanism is specified by a Layer-6 licensing protocol · back to AI access economics · audit your site.