A Media Origin Receipt™ is an independent, signed, ledger-anchored record of where a piece of media came from: the model, the prompt, the source assets, the edits, the credentials, the attestations, and the rights that traveled with it.
It covers human, AI, and hybrid media in one primitive. It is not a detector's opinion. It is not self-declared metadata anyone can strip. It's a receipt that you, and the people who receive your media, can check against an independent ledger.
Each existing approach answers a narrower question, and each has a failure mode that shows up exactly when it matters. A Media Origin Receipt does not replace any single signal. It is the independent record that carries all of them, plus the claims and attestations, into something a third party can verify.
| Detection / classifier | Metadata (EXIF, self-declared) | Watermark | Media Origin Receipt™ | |
|---|---|---|---|---|
| What it is | A model guessing "AI or not" from the pixels. | Fields the creator or tool wrote into the file. | A pattern the generator embedded in the output. | A signed record of the origin evidence and claims that came with the asset. |
| Independent of the creator? | Partly, but it is still a guess. | No. It's the seller vouching for the seller. | Only if the issuer is trusted & intact. | Yes. It's independently anchored on the Hive Ledger. |
| Survives editing / re-encoding? | Degrades fast with edits. | No. It can be stripped in seconds. | Often broken by crops, filters, re-encodes. | Records the edit chain instead of relying on survival. |
| Covers human + AI + hybrid? | No. It's a plain yes-or-no AI framing. | Whatever was declared, if anything. | No. It only works on the generator side. | Yes. One primitive covers all three. |
| Verifiable by a third party? | No. It can't be reproduced. | Only as far as you trust the writer. | Needs the issuer's detector. | Yes. Check the receipt against the ledger. |
| Failure mode | False positives / negatives, no audit trail. | Silent, undetectable removal or forgery. | Absent, stripped, or unreadable downstream. | Honest gaps: it says what evidence is missing. |
Pick the media type, state the origin claim, and toggle the evidence you actually have. The receipt below classifies the origin, grades how independently the evidence can be checked, and names the proof gaps, worked out directly from your inputs.
Each card loads a worked example into the builder above, with modality, origin claim, evidence set, rights, and dispute state pre-filled, so you can see the receipt M.O.R. would record for that case. Click a card, then read the classification, gaps, and JSON it produces.
The receipt backend is live today. Health, a signed sample receipt, and the M.O.R. issue and verify routes all return real, Ed25519-signed envelopes right now. Pick a call, copy it, run it.
# 1) issue a signed Media Origin Receipt (no auth)
RCPT=$(curl -sS -X POST https://receipts.thehiveryiq.com/v1/media-origin/receipt \
-H 'Content-Type: application/json' \
-d '{
"cryptographic_hash": "sha256:2c26b46b68ffc68ff99b453c1d30413413422d706483bfa0f98a5e886266e7ae",
"perceptual_hash": "phash:ff00ee11",
"evidence": {
"model_attestations": [{"model": "gpt-image-1"}],
"watermark_signals": [{"score": 0.91}],
"c2pa_refs": [{"ref": "c2pa:manifest-id"}]
},
"rights_state": {"asserted_scope": ["owned"]}
}')
echo "$RCPT" | python3 -m json.tool # → receipt_id, classification, Ed25519 signature
# 2) verify the exact receipt you just issued (no auth)
curl -sS -X POST https://receipts.thehiveryiq.com/v1/media-origin/verify \
-H 'Content-Type: application/json' \
-d "$(python3 -c 'import json,sys;print(json.dumps({"receipt":json.load(sys.stdin)}))' <<<"$RCPT")"
# → {"verified": true, "proof_state": "VERIFIED", "classification": "ai-origin"}
/health, and the mor_smoke.sh pilot script. That's everything an engineering team needs to evaluate M.O.R. against its own assets. Keyed / metered (fast-follow): paid /v1/receipt/emit, activation packets for fleet-scale usage, and ledger-anchored production receipts. Start free, and add a key only when you go to production.curl -sS https://receipts.thehiveryiq.com/health
A smoke run against the live endpoints, including the M.O.R. issue route POST /v1/media-origin/receipt and verify route POST /v1/media-origin/verify. These are wall-clock round-trips from a single client over 100 iterations, not a production SLA, but every number below is measured, not invented.
bash scripts/mor_smoke.sh for fresh numbers.M.O.R. works across the board. It reads whatever origin evidence a media pipeline already produces. What changes is who is asking for proof on the other side.
The cheapest moment to capture origin is when the media is made. M.O.R. binds the evidence then, so provenance is a lookup later, not a forensic reconstruction.
A receipt that names its own gaps is worth more than a detector that projects false confidence. M.O.R. grades evidence and lists what is missing.
Image, video, audio, 3D, sensor, and publication assets all resolve to the same signed receipt, checkable against one independent ledger.
M.O.R. invents no new trust. It records the origin evidence a media pipeline produces and anchors it in the same proof layer as the rest of Hive.
Explore the stack: OriginProof (human-origin attestation), Hive Ledger, Receipt Relay, R3Pv benchmark, Protected Flow Fleets, Proof Credit, media provenance, and Assurance. Verify any receipt at /verify/. Read the thinking: human-origin attestation and logs vs. receipts.
Media Origin Receipt™ and M.O.R.™ are trademarks of The Hive. Patent(s) pending.