Trust center

What a receipt is

A signed, verifiable record of what was licensed, from whom, under which terms, when. It can be used to answer provenance questions from customers, auditors and vendor questionnaires.

Every licensed retrieval produces a signed receipt, or nothing is served. The free sample documents are the one exception: they are served without metering and without receipts, and the site says so wherever it shows them.

A receipt can be verified offline with a standard JWS library and our published keys.

A receipt never states what the buyer paid, their plan, or their usage.

A signing key is never removed: a retired key stays published, for at least seven years, so that old receipts keep verifying.

What a publisher controls

A publisher decides what may be used, for which purpose and at what list price — for the whole title, a source, a section or a single document; the most specific rule wins, and nothing is served before an agreement is signed.

We never serve content from a publisher that has not signed.

A publisher's share is fixed when they sign and stays for the life of that agreement.

What a buyer is charged

The spend cap is a ceiling on the whole monthly bill.

When a request would take the month's bill past the spend cap it is refused — with a machine-readable reason — rather than charged.

Each API key can carry its own monthly spend cap, which counts that key's metered spend only.

Each plan includes a number of searches a month; beyond them a search is charged at the search price, and searches count towards the same monthly cap as everything else.

Included documents are counted per calendar month and unused ones do not roll over. Training use is not licensed.

Free-plan retrievals are metered and receipted like any other, and one publisher's monthly Free allowance is shared by all Free accounts.

Our crawler

Our crawler identifies itself: a User-Agent naming us and pointing to a page about it, and a signature on every request (Web Bot Auth, RFC 9421) that can be checked against published keys. It honours robots.txt, and a named-AI-crawler disallow is recorded as the publisher's stated intent.

Our crawler reads only the feeds, sitemaps and article pages of publishers who registered them with us.

If a site's robots.txt disallows named AI crawlers, we record that as its stated intent; nothing from any publisher is served before that publisher has signed an agreement.

The sample publisher

The sample documents come from a publisher we operate ourselves: US federal press releases. They are served without metering and without receipts so that anyone can inspect the format, and they cannot be retrieved through the licensed API.

Checking a receipt yourself

You do not have to trust this page. Paste a receipt into the verifier — it runs in your browser, against our published keys.

The in-browser verifier checks a pasted receipt in the browser, against our published keys and no others: it never fetches an address found inside the pasted text, and what is pasted is not sent to us. It reports the specimen as a specimen.