Agents now read the web to search, to answer questions and to train models. FORA turns your pages into signed offers an agent can find, read, pay for and keep a receipt of. Your pages, your prices, your terms.
Most publishers can only block AI crawlers or sign a one-off deal with one AI company. A search index, an AI answer and a training run are different uses, worth different prices, and robots.txt can only say yes or no. FORA gives each use its own price and terms.
An answer agent needs today's article to answer a question, and credits you as your terms require.
A research agent reads dozens of sources for one report and needs a license for each one it uses.
A lab that wants your archive for training asks for that use. It is off unless you offer it.
Search and citation, input to an AI answer, answers for your subscribers, training: each becomes a signed offer that says who may buy, what the agent may do, for how long and at what price, with the obligations that come with it. The agent still receives your full license text.
| Search and citation | AI answer input | AI answers, subscribers | AI training | |
|---|---|---|---|---|
| How it is paid | Free | Per fetch | Subscription | Not offered |
| Who may buy | Any identified agent | Any identified agent | Agents of your subscribers | — |
| Allowed | Search index, citation | Input to an AI answer | Input to an AI answer | — |
| Not allowed | AI training | AI training | AI training | — |
| Obligations | Attribution | Attribution; usage report | Attribution; usage report | — |
You set your own uses and prices. A subscription is a term that only your subscribers' agents see; the fee is agreed outside the offer.
A small function on your CDN stands at the door of the paths you choose to sell.
// the body, for agents that read JSON { "error": "Licensed content. Negotiate access via the Exchange.", "protocol": "FORA", "fora_json_url": "https://publisher.example/.well-known/fora.json" }
The Movement for an Open Web's Search Only Terms Contract keeps search free and puts the fee you set on every other access. It covers your open paths and the scrapers that hide what they are. An agent that wants more than search buys a license through FORA.
tdl line in the User-agent: * group tells every crawler where the contract is.The agent finds the Exchange that sells your content: from your site, or through a Broker's search across catalogs.
It asks for the content and gets signed offers: the price, the permitted uses, the obligations.
It commits to one offer, pays through the Exchange and gets a signed receipt.
It fetches the file through a short-lived link bound to its identity.
It reports what it used, when the offer asks for a report.
If the file is wrong or never arrives, it opens a dispute backed by signed evidence.
Paste the address of one page and see what an AI agent gets for it: the content, your terms and a price. No account needed.
Try your website →Name your business, add your domain and set your terms. Then serve one file on your site and install one package on your CDN. The operator imports your pages from your sitemap.
Read the publisher guide →The console lists every sale with its signed offer and acceptance, which you can verify in your browser. It totals your income by day and exports it as CSV.
After setup →FORA reads the licensing signals you may already publish, identifies agents with Web Bot Auth, and imports your pages from your sitemap.
No, and neither does your online store. It makes sure that when an agent comes looking, it can find you, understand your terms and pay, instead of hitting a 403 or scraping you.
For people, nothing. On the paths you protect, an AI bot without a license gets a 403 that tells it where to buy access, and an agent that paid gets the page through a signed link. Pages you leave open stay open.
No. You choose the paths FORA protects, and every other path stays as it is. Your CDN's own bot rules keep working beside it.
Refuse AI training in the form each AI company honors, keep the search crawlers that bring your traffic, and put a contract and a price on the rest. Every bot, and a generator for your robots.txt.
Not out of AI Overviews, Copilot or Siri. They draw on Googlebot, Bingbot and Applebot, the same crawlers as ordinary search, and FORA lets them through. ChatGPT, Claude, Perplexity and Meta AI each run a search crawler of their own. On your protected paths, FORA refuses those by default, as it refuses training crawlers and bots that load a page live to answer a question, and you can let any of them through. The rest of your site stays open to them. Which bot is which.
Training is a separate permission, and it is off unless you offer it. A paid fetch doesn't include it.
Yes. A subscription is a term that only your subscribers' agents see, and its fee stays agreed outside the offer.
The protocol and FORA Server are free and open source. Run it yourself, or have it hosted for you.
See what an AI agent gets for one of your pages.