Every AI crawler that visits a website reads a file that says almost nothing: robots.txt can allow a bot or block it, and that is all. Really Simple Licensing (RSL) replaces that yes-or-no signal with a vocabulary for rights, so a publisher can say which AI uses are permitted, which are not, and what the permitted ones cost. The terms sit in a file the crawler can parse, which means they can be applied across millions of pages without a negotiation for each one.
The standard launched on 10 September 2025 and its Technical Steering Committee announced the official 1.0 specification on 10 December 2025. It matters because it is the first widely backed attempt to turn AI scraping from a fight into a transaction. The honest status as of 30 October 2026 is that the publisher side has scaled and the buyer side has not appeared, and that gap decides what RSL is worth to you.
What is Really Simple Licensing (RSL)?
RSL is an open, XML-based standard that lets publishers and platforms define machine-readable usage, licensing and legal terms for digital content. It is built on RSS, the syndication format that distributed blog and news content across the early web, and it is managed by the RSL Collective, a non-profit co-founded by RSS co-creator Eckart Walther and former Ask.com and IAC Publishing chief executive Doug Leeds.
The Collective has two jobs. It maintains the standard, and it acts as a collective management organisation: Reuters Institute reporting from May 2026 describes it as offering AI companies a blanket licence to members' content for a single price, with publishers joining by declaring it in their RSL file. Participation is non-exclusive, so joining does not stop a publisher from signing its own deals.
How does RSL work?
A publisher adds licensing terms to the files crawlers already read, principally robots.txt, and points to an RSL licence document that sets out the conditions in full. Instead of a line that simply allows or disallows a bot, the publisher attaches terms to the permission.
RSL 1.0 introduced AI-specific usage categories - "ai-all", "ai-input" and "ai-index" - which give finer control than a blanket block. A publisher can, for example, let a search engine list its pages while opting out of AI search applications. A new "contribution" option lets creators and non-profits ask for a monetary or in-kind contribution from AI systems that benefit from their noncommercial work, developed with Creative Commons.
The payment models run from free use and attribution through subscription to pay-per-crawl and pay-per-inference. Pay-per-crawl charges each time an AI application fetches the content. Pay-per-inference charges each time the content is used to produce a response, which ties payment to genuine use rather than to access. Leeds has said the Collective's revenue model will likely take a percentage of revenue with minimum thresholds, and that a royalty would be generated for each AI response that draws on a member's work, but no rates have been published.
The published RSL specification also describes supporting protocols for acquiring licences and for crawlers to prove they hold one. I did not re-verify their details against the specification this month, so treat the protocol names in older write-ups as something to check against rslstandard.org.
Who supports RSL?
Launch supporters included Reddit, Yahoo, People Inc, Ziff Davis, Quora, Medium and O'Reilly Media, with Fastly as an infrastructure partner. Arena Group, BuzzFeed, USA Today Co and Vox Media joined in November 2025, when Digiday reported more than 50 partners. By December 2025 Leeds told Press Gazette that about 1,500 media organisations endorsed the standard, including People Inc, Yahoo and the Associated Press, and Reuters Institute repeated the roughly 1,500 figure in May 2026, adding the Guardian to its list of named members.
Two cautions apply to that number. It is self-reported by the Collective, and the Press Gazette interview does not explain what an endorsement commits an organisation to. Support is also spreading beyond the English-speaking market: in January 2026 Spain's reprographic rights organisation CEDRO became the Collective's exclusive partner in Spain, taking a seat on its technical committee.
On the infrastructure side, Cloudflare's Will Allen said a standard licence could be included in the HTTP 402 payment-request responses Cloudflare already supports, and Akamai's Eric Graham called RSL meaningful momentum toward publisher controls. Supertab runs a managed RSL server, and WordPress plugins exist to generate the files.
How is RSL enforced?
The standard states terms. It does not collect money, and the Collective's own founder has said so: Leeds told Digiday that RSL has no built-in enforcement and that enforcement would come from CDNs, lawsuits and government bodies.
Infrastructure partners provide the teeth where they apply. Fastly describes itself as the bouncer at the door, checking whether a crawler has agreed to a licence before serving content, which works for publishers served through Fastly. Cloudflare's pay-per-crawl uses the HTTP 402 "Payment Required" response so a publisher can set a price that the crawler either pays or walks away from. A publisher without an enforcing CDN can publish RSL terms but has no technical way to compel compliance.
Leeds has also argued that the file has legal value, saying that courts and regulators look to industry norms when assessing reasonable notice, access and compliance. That argument depends on RSL becoming a recognised norm, and nothing in the sources I read shows a court relying on an RSL file yet.
Will AI companies actually pay?
So far no AI company has been named as a licensee. The RSL website lists publishers, CDNs and Creative Commons as featured supporters and reports no signed deals with AI firms. Reuters Institute, on 19 May 2026, named no AI company as a licensee and said the Collective's licensing plan was still being reviewed by member organisations and was not yet active.
That sequencing was deliberate. Leeds said in November 2025 that the goal was to build supply-side scale before approaching the demand side. It is a sensible strategy for a collective, and it also means that every month without a buyer is a month in which RSL is a price list with no customers.
There are structural obstacles beyond reluctance. Pay-per-inference needs the AI system to log which sources shaped an answer, and few systems expose that. There is also a record of evasion: Cloudflare has reported stealth crawling that routes around restrictions, which shows that some operators treat access controls as obstacles. The likeliest route to compliance may be legal rather than voluntary, since AI copyright cases still in the courts could make paying for content the cheapest option available.
How is RSL different from llms.txt and licensing deals?
RSL is not llms.txt. That file helps an AI system find and understand your content and says nothing about payment, whereas RSL is about compensation, not comprehension. The two can sit on the same site.
It is also not a bespoke licensing deal. The headline agreements between AI companies and individual publishers are privately negotiated and limited to a few large names. RSL's aim is to give a mid-sized or small publisher the same machine-readable terms a giant would have, and to let the Collective negotiate in bulk. Whether buyers turn up is the open question. For how bilateral deals and collectives compare in practice, see the related pages on AI licensing deals and licensing marketplaces.
What does RSL mean for publishers?
RSL is worth adding as a low-cost option. It declares a price, it documents your terms if a dispute later turns on notice, and on an enforcing CDN it can begin working today. The realistic expectation is an option you take out now rather than a revenue line you can bank on this year.
It is also worth being clear about what RSL governs. It is a control on the input: whether a crawler may read and use your content, and on what terms. It does nothing about the output, the moment an AI assistant or live search agent surfaces an answer for a user and a brand could appear in it. A publisher can act on both layers, licensing the read where it is enforceable and capturing the answer through contextual advertising in AI responses. blankspace works at that second layer, detecting live search agent traffic at the CDN edge and injecting contextual brand mentions into AI answers. The two approaches are complementary rather than competing.
Frequently asked questions
What does RSL stand for?
RSL stands for Really Simple Licensing. The name echoes RSS, Really Simple Syndication, and the standard is co-created by an RSS co-creator. The intent is the same: take a practice that needed custom work and express it in a simple, open, machine-readable format that any site can adopt.
Is RSL free to use?
Yes. RSL is an open standard, and the RSL Collective invites publishers to join for free. Adding terms costs nothing in itself. The cost sits in enforcement and collection, which in practice depend on a CDN partner such as Fastly or Cloudflare, and the Collective is expected to take a share of any royalties it collects.
What is the difference between pay-per-crawl and pay-per-inference?
Pay-per-crawl charges an AI company each time it fetches your content, so payment is tied to access. Pay-per-inference charges each time the AI uses your content to generate a response, so payment is tied to actual use. Publishers tend to prefer the second because it reflects value delivered, but it is harder to enforce because the AI system has to log which sources shaped each output.
Have any AI companies agreed to pay through RSL?
None has been named as a licensee in the sources I reviewed. The RSL website reports no signed deals with AI firms, and Reuters Institute reported in May 2026 that the Collective's licensing plan was not yet active. Support so far comes from publishers, CDNs and standards bodies, which is the central limitation of the standard today.
Should publishers add RSL to their site now?
For most publishers it is a reasonable, low-cost step. It declares your terms, creates a record that could matter in future disputes, and positions you to collect if buyers or the courts move towards paid access. Treat it as option value rather than revenue, and pair it with a plan for monetising the AI answers your content already appears in, since RSL on its own only addresses the read.