What Is RSL?
Really Simple Licensing (RSL) is an open web standard that lets a publisher state, in a machine-readable file, how AI systems may use its content and what they owe for it. It launched on September 10, 2025, led by Eckart Walther, a co-creator of RSS, and Doug Leeds, a former Ask.com chief executive, through the nonprofit RSL Collective. Reddit, Yahoo, and Medium were among the first to back it, and version 1.0 was later published as an official specification. Where robots.txt can only say yes or no to a crawler, RSL adds terms.
How RSL Works
A publisher writes an RSL document in XML and points to it. The 1.0 specification defines four ways to do that:
- robots.txt: a
License:line with the URL of the RSL document. - HTTP header: a Link header with
rel="license"and the typeapplication/rsl+xml. - HTML: a link or script element in the page.
- RSS: licensing elements inside feed items.
License: https://example.com/rsl.xmlInside the document, permits and prohibits elements list allowed and disallowed uses, and a payment element sets the compensation.
| Usage value | Meaning |
|---|---|
| ai-train | Model training or fine-tuning |
| ai-input | Retrieval-augmented generation or grounding |
| ai-index | Inclusion in an AI system's index |
| ai-all | Any use by AI systems |
| search | Search indexing and results |
| all | Any automated processing |
| Payment type | Meaning |
|---|---|
| purchase, subscription | Direct monetary compensation |
| training, crawl, use | Fees per activity |
| contribution | Good-faith support, for example for non-commercial content |
| attribution | Visible credit is required |
| free | No payment needed |
Declaring Terms Versus Enforcing Them
An RSL file is a statement of terms. By itself it does not stop a crawler. The specification adds companion protocols for enforcement: the Open License Protocol, an OAuth 2.0 extension through which a crawler obtains a licence token, the Crawler Authorization Protocol, which sends that token in an HTTP Authorization header so a server can check it at request time, and the Encrypted Media Standard for content that is only decryptable by licensed clients. In practice enforcement depends on a CDN or server refusing unlicensed requests, and on AI companies choosing to read the file.
Why It Matters for AI Visibility
RSL gives a site a third option between allowing everything and blocking everything. The useful distinction is between ai-train and ai-input. A brand that wants to be cited in AI answers but does not want to feed training can prohibit one and permit the other. Getting this wrong has a cost: prohibiting ai-input or search tells compliant assistants not to retrieve your pages, which removes you from their answers. Publishers are the main audience for RSL. Most brands selling products will want retrieval allowed.
Commonly Confused With
RSL is not llms.txt, which is a content map that helps models read a site and says nothing about rights. It is not pay per crawl, which is a way of charging at the CDN: RSL can describe a per-crawl fee, and a pay-per-crawl system can collect it. It also differs from marketplaces such as TollBit, compared in TollBit vs Cloudflare Pay Per Crawl, which broker deals rather than define a file format.