Usage reporting: what AI did with your content
A label shows where content came from. A usage report says what an AI system did with it: fetched it, used it in an answer, cited it, showed it, or sent a click. Your logs only see the first.
- The standard
- The Content Telemetry StandardContent Telemetry Standard: An open format for AI companies to report what they did with content: fetched, used, cited, shown or clicked. from the SPUR Coalition of news publishers.
- Version 1.0
- Published on September 2, 2026, under the Apache 2.0 license.
- Who reports
- You see fetches. Only the AI system can report the rest.
- Encypher's part
- We took part in the public consultation. SPUR runs the standard.
The five things an AI system can report
A reader asks a chatbot a question. The answer draws on your story and adds a link nobody clicks. Your logs show one fetch. The standard names five events. The SPUR Coalition launched in February 2026; its founding members include the BBC, the FT and the AP.
| Event | What happened | Who can see it |
|---|---|---|
| Retrieved | A bot fetched the content. | You, in your own server logs |
| Grounded | The content went into what the AI used to write an answer. | Only the AI system |
| Cited | The answer named the source. | Only the AI system |
| Presented | The content or a link to it was shown or read aloud. | Only the AI system |
| Engaged | A user or agent acted on it, such as with a click. | The AI system, or the site the click lands on |
Who reports what
Your server or content network sees fetches, and a destination site can confirm a click. Only the AI system can report grounding, citing and showing. Each AI report is a claim by a firm that may owe you money, so ask for audit rights.
Where C2PA labels fit
A web address changes when a story is syndicated or cached. A C2PA manifestC2PA manifest: The signed label itself: a small package inside a file that records who signed it, what they declared and how to check it. gives the report a stable content ID instead. Encypher labels also carry your license terms and a usage-reporting request. A label cannot enforce either one. A license contract can.
What reports can and cannot show
| Reports can show | Reports cannot show | |
|---|---|---|
| Use | What the AI firm says it fetched, used, cited and showed. | That the firm reported every use. |
| Credit | How often your work went into answers without credit. | How much your work changed an answer. |
| Money | One format, so reports from many firms add up. | What anyone owes. The contract decides that. |
| Training | Use at answer time. | Model training, which is outside the standard. |
What to do now
Sign when you publish.
Put your story ID, license terms and a reporting request in each label. Check that a published copy keeps it.
Count AI fetches yourself.
Your logs give a number to check reports against. SPUR has free code for Cloudflare and Fastly.
Write reporting into every AI deal.
Name the events, where they go, and your audit rights. An RSL (Really Simple LicensingRSL: Really Simple Licensing: a standard for publishing AI licensing terms that robots.txt can point to.) license can require reports, and IAB CoMP can say where to send them. Set your terms.
For engineers
Show the technical detail
- Matching
- A shared Content-Telemetry-ID header lets a publisher's fetch log and the AI firm's report be joined on the same fetch. Redirects can drop it.
- Content IDs
- CTS accepts a C2PA manifest hash as a stable content identifier, and a grounding event can report a detected provenance signal.
- Identifier prefixes
- Publishers register an ID prefix with the collector, so reports keyed on that ID route without a URL.
- Coverage
- Report signing is optional in v1.0. Absence means nothing unless the firm declares it reports every qualifying use.
Create. Mark. Endure.
Sign your stories with your own ID and license terms, so reports can find their way back to you.