Straight answer: RSL, which stands for Really Simple Licensing, is a machine-readable standard that lets a website declare what AI companies are and aren’t allowed to do with its content, and what they owe you if they use it anyway. It’s built by a group backed by Reddit, Yahoo, Medium, O’Reilly and a few others, and it works a bit like robots.txt with a price tag attached. For most small businesses, adding an RSL license file is worth an afternoon, but expecting it to bring in money is, honestly, wishful thinking right now.
More on this here: AI News This Week for Small Business: 6 September 2026.
What RSL is, in plain English
For years, robots.txt has told search engines and bots what they can crawl. It’s a blunt instrument. It says yes or no, it doesn’t say “yes but only if you pay me” or “yes but you must credit me” or “no training my content into your model, full stop.” RSL fills that gap.
The RSL Collective launched the standard in September 2025, with Reddit, Yahoo, Medium, O’Reilly, Quora, and wikiHow among the early names attached to it. The idea is simple: a website adds a license file, written in XML, that sits alongside or references from robots.txt, and it spells out terms an AI crawler is meant to respect. Free with attribution. Paid per crawl. Paid per token used in training. No training allowed at all, only inference. Subscription access. It’s the same logic as a Creative Commons license, except it’s aimed squarely at large language models hoovering up the internet to train and to answer queries.
The bit people miss is that RSL is a declaration, not an enforcement mechanism. It’s you putting a sign on your gate. Whether anyone reads the sign, let alone obeys it, is a separate question entirely.
Why this exists now
Publishers spent 2023 and 2024 watching their traffic get quietly cannibalised. ChatGPT, Perplexity, Google’s AI Overviews and Claude were all trained on, or actively pulling from, content that nobody paid a license fee for. Some big players fought back with lawsuits. The New York Times sued OpenAI and Microsoft in December 2023 over exactly this. Others, like News Corp and the Associated Press, cut direct licensing deals worth tens of millions of dollars.
Small businesses don’t get lawsuits or seven figure deals. What we got, eventually, was a standard we can bolt onto our own sites for free, so at minimum there’s a documented, timestamped record of our terms if this ever becomes legally meaningful, which it might. Courts have already started treating clear licensing terms as evidence of intent in copyright disputes.
A quick story from my own site
Around mid 2025 I noticed a client’s blog, a niche accountancy firm with maybe 40 solid articles on VAT and payroll, had a strange pattern in its server logs. Bot traffic from a handful of AI crawler user agents (GPTBot, ClaudeBot, one labelled PerplexityBot) had gone from almost nothing to being a meaningful chunk of total requests, more hits some weeks than actual human visitors reading the blog. The content was being pulled and, we assume, fed into answers on other platforms, while the firm’s own organic traffic to those same pages had slipped by around 18 percent over the previous year.
We didn’t sue anyone, because that’s absurd for a firm that size. What we did was add a robots.txt directive blocking the crawlers we could identify, and a few months later, once RSL matured a bit, we added a license file setting terms of “free to summarise with attribution and a link back, no verbatim training use.” Did it stop the scraping entirely? No. Did it change how the AI companies behaved? Almost certainly not, because most of them aren’t checking RSL files yet and there’s no legal requirement that they do. What it did give the firm was a clean, dated record that they’d set explicit terms, which their solicitor said was worth having on file “just in case.” That’s the honest scope of the win.
How to add an RSL license (step by step)
If you decide it’s worth doing, here’s the practical version, not the theoretical one:
- Write your license file as XML, following the RSL schema (the RSL Collective publishes the spec and examples on rslstandard.org, and it’s short, most sites need under 30 lines).
- Decide your terms: attribution required, no training allowed, pay per crawl, or open access. Most small sites go with “attribution required, no verbatim training reuse” because it’s simple and doesn’t need a payment system behind it.
- Host the file at a predictable location, usually /rsl.xml or referenced from your robots.txt with a Content License line pointing to it.
- Update robots.txt to reference the RSL file and to explicitly allow or block the named AI crawlers you care about (GPTBot, ClaudeBot, Google-Extended, PerplexityBot, CCBot are the common ones as of now).
- Check it validates. There are free validator tools listed on the RSL Collective’s own site, so you’re not guessing whether the syntax is right.
- Revisit it every few months, because the list of crawler user agents and the standard itself are both still moving targets in 2026.
That whole process takes an evening for someone comfortable editing a text file, or an hour of a developer’s time if you’d rather pay for it and never think about XML again.
The uncomfortable bit nobody selling this standard wants to say
Here’s what most of the coverage on RSL skips over. The reason Reddit, Yahoo and Medium can make this work is that they have scale, legal teams, and direct relationships with AI companies who want their content badly enough to negotiate. A small business blog with 40 articles and 2,000 monthly visitors has none of that use. If OpenAI or Google decides your RSL terms are inconvenient, the realistic outcome isn’t a payment negotiation, it’s that your content either keeps getting pulled anyway or gets quietly deprioritised in whatever crawl queue exists. There is currently no independent enforcement body checking compliance, no fine, no automatic royalty system that pays out to a solo consultant the way ASCAP pays out to a songwriter when a song gets played on the radio. So if anyone tells you adding an RSL file will start generating licensing income for your small business blog, they’re selling you something, or they haven’t thought it through. The honest version is that RSL right now is a governance and record-keeping tool, not a revenue stream, for anyone below enterprise scale. That may well change, several people involved in the standard have talked openly about building a marketplace layer on top of it, similar to how Cloudflare has started brokering pay-per-crawl deals for publishers on its network. But “may change” and “works today” are different claims, and small business owners deserve the difference stated plainly.
Should your small business bother with it
My honest recommendation splits by situation:
- If your content is original research, data, or reporting that AI models would want to train on directly (surveys, proprietary data, long form investigative pieces), add the license file. It costs you almost nothing and it puts a legal marker down for later.
- If your content is standard service-business writing, how-to guides, FAQ pages, product descriptions, it’s still worth a robots.txt update to block the crawlers that don’t respect anything at all, but don’t expect the RSL layer to change your bottom line this year.
- If you’re running the kind of small business where content is a genuine side revenue stream, similar to some of the side hustle ideas busy business owners lean on to diversify income, keep an eye on Cloudflare’s pay-per-crawl program and similar marketplaces rather than the raw RSL standard alone, because that’s where actual payment mechanisms are starting to appear.
- If you don’t have the technical resource in house and you’re already thinking about how AI touches your wider marketing stack, this is exactly the kind of decision that sits well within what a fractional AI officer for a small business would flag as part of a quarterly AI audit, rather than something you bolt on reactively.
I’d also say this plainly: don’t let RSL become a distraction from bigger AI decisions your business needs to make this year, like whether you’re using AI image tools safely (worth reading before you touch a free AI image generator for client work, and worth a proper look if you’re trying to compare AI image generators for business use), or how you’re pricing AI help if you bring someone in. If that’s on your radar, it’s worth understanding roughly what an AI consultant typically costs before you commission anything RSL-related as a standalone project.
What I’d tell a client this week
If a client asked me today, I’d say: spend the hour, add the file, block the crawlers you don’t want scraping you for free, and move on. Don’t pay a developer 500 pounds to build something fancy around it. Don’t hold your breath for royalty cheques. Treat it the way Dan Kennedy would have treated any new marketing gadget, and there’s a reason his approach still gets referenced in the business lessons from Dan Kennedy people keep coming back to: test it cheap, measure what it returns, and don’t fall in love with a tactic just because it’s new and everyone’s talking about it. RSL is a sensible, low cost thing to have in place. It is not, yet, a business model.
Frequently asked questions
Does adding an RSL license file stop AI companies from scraping my site?
No, not on its own. It’s a declared license, similar to robots.txt, and there’s currently no legal or technical mechanism forcing AI crawlers to obey it, though some do respect it voluntarily and it strengthens your position if a dispute ever goes legal.
Is RSL licensing free to set up?
Yes, the standard itself is open and free, you’re just writing an XML file and updating your robots.txt. The only cost is time, or a developer’s hourly rate if you outsource it, typically under an hour of work.
Can a small business earn money from RSL licensing?
Not meaningfully yet. Payment mechanisms are mostly working for large publishers with direct deals or through emerging marketplaces like Cloudflare’s pay-per-crawl program, not for individual small business sites, so treat any income as speculative for now rather than expected.
Should I block AI crawlers entirely instead of licensing them?
That’s a valid choice, and for content you don’t want summarised or trained on at all, blocking via robots.txt is more reliable than hoping a license file gets respected, though it also means you lose any chance of appearing in AI-generated answers that could send you referral traffic.
