Victor Kuznetsov
304bd071c0
Share producer identities across mark routing, attribution, and pinned signers. Preserve nested Honor TC260 fields and report unsupported RSA signatures. Document the Google Photos checks and public sample search. Use an SDK-signed synthetic carrier for negative EKU checks after the original Microsoft certificates expired. Skip boundary regex scans when no C2PA soft-binding identifier is present.
Provenance fixtures
These files exercise real container, metadata, and pixel-provenance formats. They are test inputs, not evaluation corpora.
Honor YOYO metadata reconstruction
synthetic/honor-yoyo-tc260.json reconstructs the EXIF UserComment JSON
structure observed in two public YOYO edits in this
Honor Club thread, fetched
on 2026-10-01 UTC. The downloaded JPEGs have SHA-256
99aa5cdf42cd4c3849401a32fc1f0738c229cc3ed903eba00a080e7eb29542d5 and
af779adba954926bb15bb9cb70ff597cb11e715b53e204fce5e64b1edf971aa1.
They remain untracked because public redistribution has not been cleared.
The fixture retains the observed producer 001191440300MA5G49LC9K1YO01,
nested ReservedCode1/ReservedCode2 objects, PubSd spelling, null hash
binding, and RSA/SHA-256 algorithm identifiers. Generation IDs, signature,
and public key are explicit synthetic placeholders. It contains no photograph,
account handle, real signature, or claim of a valid asset binding. The producer's
USCC matches Honor Device Co., Ltd. in its
official privacy statement.
tests/test_web_sample_fixtures.py is the carrier recipe: it embeds this JSON
in a solid-color JPEG, checks Honor attribution and preservation of nested JSON,
reports the content signature as unsupported, and verifies removal preserves
decoded pixels. This synthetic test input
is distributed under the repository license. It does not establish gallery
control/crop behavior, device/app versions, or a real verified RSA signature.
OpenRouter metadata survey
The following files are original byte streams returned by OpenRouter's Image
API on 2026-09-02. Each request used a repository-maintainer account, the same
neutral still-life prompt, one 1:1 output, and no image reference. The prompt
was A documentary photograph of three smooth river stones on pale linen beside a small blue glass vase, soft natural daylight, plain uncluttered background, no text, no logos.
| File | Model | Cost | SHA-256 | Expected metadata result |
|---|---|---|---|---|
recraft-v3.webp |
recraft/recraft-v3 |
USD 0.04 | e94c99ab50410092dbb601c8269ce812879e2271e18c7880021d2118666542ae |
Valid C2PA; recraft.ai; trainedAlgorithmicMedia |
krea-2-medium-turbo.png |
krea/krea-2-medium-turbo |
USD 0.015 | ee1302c01e8da8e316e6563b0f9d36d511b9048fb2f70b8851eeeda98c5fb6a0 |
No locally readable provenance |
Both were paid API generations. Recraft grants ownership for paid-plan output, and Krea states that users own compliant outputs. They are distributed under the repository license.
- Recraft terms: https://www.recraft.ai/docs/plans-and-billing/commercial-rights-and-ownership
- Krea terms: https://www.krea.ai/krea-2-licensing
The same survey generated sourceful/riverflow-v2-fast output and found no
locally readable metadata. It is not committed because Riverflow's terms
restrict redistribution and using results to test or improve machine-learning
systems. Its measured SHA-256 was
d4d7e80a3ed450129a0e77bdf330a83eb7de1e287d5b2b737cd26a011194a270;
OpenRouter recorded a USD 0.02 request cost.
WaveSpeedAI metadata survey
The following measurements came from original byte streams returned by the WaveSpeedAI API on 2026-09-03. Each request used a repository-maintainer account and the same neutral still-life prompt recorded above. The account balance moved from USD 12.00 to USD 11.82 for the four successful requests.
| File | WaveSpeedAI model | SHA-256 | Expected metadata result |
|---|---|---|---|
qwen-image.png |
wavespeed-ai/qwen-image/text-to-image |
5b6d832834397498d0587b9b2536f88c75e923195c43c2b71d9f7f3c5422b4f1 |
No provenance signal; one non-provenance PNG job-id text field |
hunyuan-image-3.png |
wavespeed-ai/hunyuan-image-3 |
6ae2ac073ac79b02b0934c905364f200420c2ff60fea232bd9e0a5fca785cc09 |
Valid C2PA from fal.ai naming fal-ai/hunyuan-image, plus Adobe TrustMark Variant P, schema 0 |
kling-image-v3.png |
kwaivgi/kling-image-v3/text-to-image |
501ab865b47ba59b64bc0f68118300744b0d5c7be2fa4e2ecbb3a78de5d8a82b |
TC260 PNG AIGC field naming producer 001191110108335469089C10100; no visible Kling mark |
seedream-v4.jpg |
bytedance/seedream-v4 |
9ec6af1fc0509d74e264b9631a6cddd2ce7ce20146347fe418e9306032beb759 |
No locally readable provenance |
All four original byte streams are committed. Section 11.1 of WaveSpeedAI's terms assigns its rights in Customer Data, explicitly including Outputs, to the customer. The live catalog records used for these requests expose no additional license, terms URL, attribution requirement, territory restriction, or commercial-use restriction. The maintainer therefore distributes these paid, neutral-prompt outputs under the repository license. This conclusion applies to these WaveSpeedAI API transactions; it does not import unrelated terms from the vendors' direct consumer services.
- WaveSpeedAI terms: https://wavespeed.ai/static/terms
- WaveSpeedAI commercial-use policy: https://wavespeed.ai/docs/commercial-use-policy
- Qwen-Image license: https://github.com/QwenLM/Qwen-Image/blob/main/LICENSE
Qwen Create consumer output
qwen-create-qwen-image-2.png is the original 2048 x 2048 PNG downloaded from
an authenticated Qwen Create generation on 2026-09-03. It was generated by
Qwen-Image 2.0 at 1:1 using the neutral still-life prompt recorded above.
The current global consumer surface differs from the Chinese Qwen text mark: it
places a small Qwen symbol at the bottom right, not 千问AI生成. The Qwen
engine has separate text and symbol templates; the symbol path scores 0.848 on
this fixture while the text path alone scores 0.178.
- SHA-256:
60c39ed8aedb081193c62289be549785e74d31fc8d071961003fabd74712429d - Expected metadata signal: TC260 AIGC label
- Expected
ContentProducer:001191440101MA9Y9T4H7A00000 - Expected visible result: Qwen symbol detected and removable under the
qwenregistry key - License: repository license; generated and explicitly cleared by the maintainer
- Qwen Create terms: https://terms.alicdn.com/legal-agreement/terms/c_end_product_protocol/20260519200335649/20260519200335649.html
Open DWT-DCT
sdxl-dwt-dct.png is a real AI-generated carrier with the public 48-bit SDXL
watermark embedded by ShieldMnt's invisible-watermark encoder using its
dwtDct method. The carrier is the original kling-image-v3.png WaveSpeedAI
output documented above, not a procedurally generated test texture. Encoding
was lossless to PNG and removed the source container metadata; the resulting
fixture contains only the expected open DWT-DCT signal.
- Source carrier SHA-256:
501ab865b47ba59b64bc0f68118300744b0d5c7be2fa4e2ecbb3a78de5d8a82b - Fixture SHA-256:
9ec3dcbfea19e4651b898f98ada28a1c81f054d63621a0bba2d484aa77da251a - Payload:
0b101100111110110010010000011110111011000110011110 - Expected signal: open DWT-DCT,
Stable Diffusion XL - Verification: 48/48 payload bits recovered by both the upstream decoder and
the in-tree decoder;
identifyreports the expected high-confidence signal - Encoder: https://github.com/ShieldMnt/invisible-watermark
- License: repository license; the encoder is MIT-licensed
Microsoft InvisMark
Microsoft Paint
microsoft-paint-invismark.png is the original byte stream supplied directly
by Xusheng Li on 2026-09-02 after his reverse-engineering research documented
Paint's pixel watermark and its matching C2PA soft binding. He authorized public
redistribution of the fixture under the MIT license. The neutral image was
generated by Microsoft Paint from the prompt a cobalt blue circle above a tiny orange square.
- Source research: https://xusheng.dev/posts/reversing/mspaint_invisible_watermark/main/
- SHA-256:
74eda22d210d534fcb5b75bedf8acd3bd110b48a95f4b9e546cd34f546effc05 - Dimensions: 1254 x 1254 RGB PNG
- Expected signal: C2PA soft binding
com.microsoft.invismark.1 - Expected watermark ID:
a4ed421e-7c37-4409-998e-f8788c564572 - Expected platform:
Microsoft (Copilot / Designer) - Expected C2PA result: valid asset binding and signature
- License: MIT, reproduced in
../licenses/xusheng-li-paint-invismark-MIT.txt
Microsoft MAI-Image through OpenRouter
microsoft-invismark.png is the original byte stream returned by OpenRouter's
Image API for microsoft/mai-image-2.5-pro on 2026-09-02. The request used a
repository-maintainer account and asked for one 1:1 image. OpenRouter routed
the request to Microsoft Azure. Microsoft states that it does not claim
ownership of AI-service output; the maintainer distributes this output under
the repository license.
- SHA-256:
be27ac7c351e7f381dc69e140ca3611ecd9db81e40926c68050ceffdd0ce6c04 - Expected signal: C2PA soft binding
com.microsoft.invismark.1 - Expected watermark ID:
19b070a1-23f0-40eb-826c-cade306337eb - Expected platform:
Microsoft (Copilot / Designer) - Request cost recorded by OpenRouter: USD 0.110757
- License: repository license
- Terms: https://www.microsoft.com/en/servicesagreement/
Microsoft MAI-Image through Bing Image Creator
microsoft-bing-image-creator.jpg is the original download produced without
sign-in by Microsoft's public Bing Image Creator on 2026-09-03. The page named
the model MAI-Image-2.5-Flash; it used the neutral still-life prompt recorded
above and the 3:2 aspect ratio. The downloaded image has no visible AI badge,
so it does not close the separate visible Microsoft-badge fixture request.
- SHA-256:
55cad5355b6eeeb25f84ea074f85c0ea55b82d63c27f43e3429bb75f28734ac7 - Dimensions: 1248 x 832 RGB JPEG
- Expected signal: valid C2PA plus soft binding
com.microsoft.invismark.1 - Expected watermark ID:
aec9a34e-665f-4549-9c02-9d29337a5de6 - Expected platform:
Microsoft (Copilot / Designer) - License: repository license
- Terms: https://www.microsoft.com/en/servicesagreement/
Adobe TrustMark
adobe-trustmark-p.png is Adobe's official images/ghost_P.png Variant P
example from adobe/trustmark at commit
0ed40cbe8188f664fd9cbbeacd969807de27440a.
- Source:
https://raw.githubusercontent.com/adobe/trustmark/0ed40cbe8188f664fd9cbbeacd969807de27440a/images/ghost_P.png - SHA-256:
e58c5825ed7e5d9fb04710ea541b61bd55879cad65554c0d46260aa24b3d0755 - Expected signal: TrustMark Variant P, schema 1
- License: MIT, reproduced in
../licenses/adobe-trustmark-MIT.txt
Ideogram C2PA attribution
ideogram-c2pa.png is a SYNTHETIC signed manifest, not a vendor download: a
64x64 flat PNG carrying a C2PA claim (trainedAlgorithmicMedia,
claim_generator "Ideogram/3.0") signed by a throwaway test CA whose leaf
certificate subject is O=Ideogram, Inc, CN=Ideogram, Inc. It exercises the
issuer-to-platform ATTRIBUTION path end to end through the real reader; the
signer is deliberately untrusted (no trust anchors ship), which is also the
fixture's second assertion: attribution and signer trust are separate
dimensions.
- Expected signal: C2PA, platform
Ideogram,ai_from_metadataTrue - Expected signer:
untrusted(test CA; attribution must not imply trust) - Recipe (no keys are committed): create a CA and an ES256 leaf with
digitalSignaturekey usage and subjectCN=Ideogram, Inc, thenc2pa.Builder.from_json(...).sign_file(png, out, Signer.from_info( C2paSignerInfo(b"es256", leaf_pem + ca_pem, pkcs8_key, None))). - No third-party image and no Ideogram artifact contributes pixels or bytes.
Gemini API images and Gemini Omni video
Original outputs generated by the maintainer and cleared for publication. The prompts were neutral scenes (a red bicycle, a paper boat, a coffee cup).
| File | Source | Size | SHA-256 |
|---|---|---|---|
google-nano-banana.png |
Gemini API gemini-2.5-flash-image, 2026-09-24 |
1024 x 1024 PNG | 1fd7cccffb25f18e74fa5ebc639672f1361db6137e6b9e848bc33fc1f583c0fe |
google-nano-banana-2.jpg |
Gemini API gemini-3.1-flash-image, 2026-09-24 |
1408 x 768 JPEG | 977110ac48678627d907d635ac1b110b930044c350b669fb5f927a500e7d95ac |
google-nano-banana-pro.jpg |
Gemini API gemini-3-pro-image, 2026-09-24 |
1408 x 768 JPEG | 5fd14ce6d32f3806f0128f952eb30d36f7dad384a9eec728e5efbdb46ebca7db |
google-nano-banana-2-lite.jpg |
Gemini API gemini-3.1-flash-lite-image, 2026-09-24 |
1408 x 768 JPEG | df4f422b9e77d6b08d3e6d095299f0cdf6d48f6c770632bb042a6c7b09627334 |
google-gemini-omni-flash-api.mp4 |
Gemini API gemini-omni-1.1-flash, 2026-09-24 |
10 s, 1280 x 720 H.264 and AAC | fb034e94e4cbca9e0b56d82389d361fda0d6191f7e16f37ac6c55cd62a3621f7 |
google-gemini-omni-app.mp4 |
Gemini app, Omni video with the Media watermark setting on, 2026-09-23 | 10 s, 1280 x 720 H.264 and AAC | f11595c6a9c0a72d4712d228354f589078d6940e529538cb009c799d8b8410d7 |
Every file carries a valid Google C2PA manifest signed Google Media Processing Services (claim generator Google C2PA Core Generator Library) with
c2pa.created ("Created by Google Generative AI.") and a SynthID
c2pa.edited action: "Applied imperceptible SynthID watermark." on the images,
"Added imperceptible SynthID watermark" on the videos. The API outputs have no
visible mark. The app video adds the Veo-style diamond on every frame and a
third action, "Added visible watermark".
- Expected signal: C2PA AI-generated, platform
Google (Gemini / Imagen), SynthID present by Google provenance; the app video reports visible markveo(Omni reuses the Veo diamond, so its platform readsGoogle Veo) - License: repository license; generated and explicitly cleared by the maintainer
Runway Gen-4 image and Gen-4.5 video
Original task artifacts generated by the maintainer on a Runway Standard plan on 2026-09-24 and cleared for publication; neutral coffee-mug scenes.
| File | Source | Size | SHA-256 |
|---|---|---|---|
runway-gen4-image.png |
Runway Gen-4 image, 16:9 1080p | 1920 x 1080 PNG | 12262d853760b4e6dab7a896b833331ee5af43cd6e9467b899b7de76c7361d8d |
runway-gen45-video.mp4 |
Runway Gen-4.5 text-to-video, 5 s | 1280 x 720 H.264, 24 fps | 98c0040b60aad1187c45df5fd4d6e8a11762950329b1a89183d6a58d5b947ad2 |
Both carry a valid C2PA manifest from c2pa-rs, signed with issuer RUNWAY AI, INC., recording one c2pa.created action typed trainedAlgorithmicMedia with
software agent Runway Image Generation 4.0.0 or Runway Video Generation
4.5.0. Neither has a visible mark.
- Expected signal: C2PA AI-generated, platform
Runway, no SynthID claim - License: repository license; generated and explicitly cleared by the maintainer
Amazon Bedrock Nova Canvas image and Nova Reel video
Third-party sample outputs published in
aws-samples/amazon-nova-samples
at commit ad2fe2074cbc260b9f1dbcec94a4ec3561c1159a, copied unmodified.
| File | Upstream path | Size | SHA-256 |
|---|---|---|---|
amazon-bedrock-nova-canvas.png |
multimodal-embeddings/getting-started/sample_assets/images/dog.png |
1280 x 720 PNG | b2d38079222ca8499f2cbe3adca3f5521add6b383c99bf28d61021503bb3b7c1 |
amazon-bedrock-nova-reel.mp4 |
multimodal-understanding/workshop/1-getting-started-with-nova-2/video/cell-phone.mp4 |
1280 x 720 H.264, 24 fps, 6 s | 36d2d68f9be53d9f123bd16127c6fbcbd838faa57cfdb28af44a515458a3cf64 |
Both carry a valid C2PA manifest signed by Amazon Web Services, Inc. with claim
generator Amazon Bedrock, recording one c2pa.created action typed
trainedAlgorithmicMedia with software agent Nova Canvas or Nova Reel 1.1.
The image was reported as Canva before the Amazon vendor row existed: the byte
fallback found the Canva token inside Nova Canvas.
- Expected signal: C2PA AI-generated, platform
Amazon Bedrock (Nova) - License: MIT No Attribution, see
../licenses/aws-samples-amazon-nova-MIT-0.txt
Higgsfield and Dreamina downloads
Original downloads generated by the maintainer on paid Higgsfield and Dreamina plans on 2026-09-24 and cleared for publication; neutral coffee-mug scenes. Higgsfield serves its own and third-party models, so each file tests what the aggregator does to the upstream provenance.
| File | Source | Size | SHA-256 |
|---|---|---|---|
higgsfield-flux-2-pro.png |
FLUX.2 Pro on Higgsfield | 768 x 1024 PNG | 4382b0c2a5f3d2b2c52f070ba9a12ffa259b4e0bad2446a65134cd700a06cb0a |
higgsfield-gpt-image.jpg |
GPT Image on Higgsfield, JPEG download | 1024 x 1024 JPEG | 4cb06b2084f2f1907421fbab38ccc459ea2cb33859005e69c07c3ea524dc0f6d |
higgsfield-kling-o1.png |
Kling O1 image on Higgsfield | 880 x 1168 PNG | 6ad75936db0bb172423564dd24d27bddb992917263555ecb5697cd2f354deeac |
higgsfield-veo-3-1-lite.mp4 |
Veo 3.1 Lite on Higgsfield, 8 s | 1280 x 720 H.264, 24 fps | 7d4ffa63755d605538c9a73059b4f6354a57e4f21100a92497aebc3c19dc3485 |
higgsfield-kling-3-0-turbo.mp4 |
Kling 3.0 Turbo on Higgsfield, 5 s | 1280 x 720 H.264, 24 fps | 7011c92d52ec624da12600f24a4abb8d997f26cdea62d6f5a8d1940bc7ffcc9c |
higgsfield-happyhorse.mp4 |
HappyHorse on Higgsfield, 5 s | 1280 x 720 H.264, 24 fps | 967dcb6022e19f55639ff705a58e5252a62a24405f247dcffefcc0f610149c30 |
higgsfield-minimax-h3.mp4 |
MiniMax Hailuo H3 on Higgsfield, 5 s | 2560 x 1440 H.264, 24 fps | 5fe5b9d08b9fb41a4d3eea470d443eb5df520235300e65134090338701215a43 |
higgsfield-cinema-studio-v2.mp4 |
Cinema Studio Video v2 on Higgsfield, 3 s, 2026-09-25 | 1280 x 720 H.264, 24 fps | 485449b69f1013a97970aabc5429327db1ff88559c81e3430c536fab108db7dc |
higgsfield-hailuo-2-3.mp4 |
MiniMax Hailuo 2.3 on Higgsfield, 6 s, 2026-09-25 | 1366 x 768 H.264, 24 fps | 67f4b6c49f6b93485fd47c457f64d3f721ec326ec57a9fb39488d6b9023528b7 |
higgsfield-genjutsu-motion-control.mp4 |
Higgsfield Genjutsu motion control on our Wan 2.5 clip and FLUX.2 image, 2026-09-25 | 842 x 486 H.264 | a3c2e8c506327ff0bdee078372be8d89fb913672f61efc66a4791771026ec0fc |
dreamina-seedream-5-pro.jpg |
Seedream 5.0 Pro on Dreamina | 2368 x 1776 JPEG | 2393cb470dc7c69de4c421f338aeb08192ed916515bc095a3afcd2ef4af89c90 |
dreamina-seedance-2-mini.mp4 |
Seedance 2.0 Mini on Dreamina, 5 s | 1280 x 720 H.264, 60 fps | d2712d4804074ac08e28fc77ff1c1ff986e6c4bbaf3ae9737f743719e51a59de |
The FLUX.2 Pro PNG keeps Black Forest Labs' C2PA manifest, but Higgsfield
inserted its hf-job-id text chunk after signing, so validation fails with
assertion.dataHash.mismatch. The GPT Image JPEG keeps a valid OpenAI manifest
and has no hf-job-id. The Kling O1 PNG carries a TC260 label with Kuaishou's
producer code plus hf-job-id. Both the GPT Image and the Kling O1 file score
0.53 as a Gemini sparkle on wood texture. The Veo clip keeps Google's C2PA with
its SynthID action, and the Kling detector matched its wood texture on every
frame. The three Chinese videos carry only an MP4 AIGC TC260 label, with the
producer written as the bare string kling, as Alibaba Tongyi Yunqi's code
(001191330106MA2CFLDG4R10001), or as the bare string MiniMax. Dreamina files
carry a ByteDance C2PA manifest whose active claim is a transcode, with the
Dreamina/7.5.0 generator in the ingredient, and a visible boxed AI badge in
the top-left corner that the library does not detect. Higgsfield's own Cinema
Studio v2 carries Kling's TC260 label (producer kling); neither it nor the
Hailuo 2.3 clip has a visible mark, and both were false visible kling and
hailuo detections on wood grain before the swirl edge gate and the Hailuo
white-pixel gate. The Genjutsu clip has no provenance at all and was a drifting
wordmark match on wood grain until Kling runs had to stay anchored within 0.95
IoU.
- Expected signal: FLUX.2 Pro, AI at medium confidence with platform
Higgsfield (model not identified)and invalid C2PA; GPT Image, OpenAI; Kling O1, the TC260 label; neither of those two reports a Gemini sparkle. Veo,Google (Gemini / Imagen)with SynthID present and no Kling mark; Kling,Kuaishou Kling AI; HappyHorse,Alibaba Tongyi (Wan, HappyHorse); MiniMax,MiniMax Hailuo AI; both Dreamina files,ByteDance Dreamina; Cinema Studio v2 and Hailuo 2.3, no visible mark - License: repository license; generated and explicitly cleared by the maintainer
xAI Grok video and YouTube re-encodes
Original outputs and platform round trips produced by the maintainer and cleared for publication; neutral scenes (a coffee cup, a paper boat).
| File | Source | Size | SHA-256 |
|---|---|---|---|
xai-grok-imagine-video.mp4 |
Grok Imagine Quick Animate, SuperGrok, 2026-09-23 | 6 s, 448 x 672 | c6e3329009260cd230deaadd0cc7293269a94c4628010cd7eeb3d390b76cc0c2 |
youtube-reencode-grok.mp4 |
YouTube Studio download of the Grok video uploaded unlisted, 2026-09-24 | 6 s, 448 x 672 | 236dd73a25a893b8f2c7bbdad8e1208341251c2b12e7b261c1fff65bf58277a5 |
youtube-reencode-gemini-omni-flash.mp4 |
YouTube Studio download of google-gemini-omni-flash-api.mp4 uploaded unlisted, 2026-09-24 |
10 s, 1280 x 720 | aa2fde5c0444eda05f40db2a8974183284016876c9788f099dda9df5aac1714f |
The Grok original carries a C2PA manifest with claim generator Grok Imagine
and a trainedAlgorithmicMedia source, signed by a self-signed ephemeral
certificate. YouTube re-signs both uploads with a valid Google LLC manifest
from YouTube Video Processing Services recording c2pa.opened and
c2pa.transcoded. The Omni ingredient keeps its SynthID action; the Grok
ingredient's self-signed manifest is dropped by the reader.
- Expected signal: Grok original, AI with platform
xAI Grok Imagineand no SynthID; YouTube Grok re-encode, unknown with no platform and no SynthID; YouTube Omni re-encode, AI with SynthID present by Google provenance - License: repository license; generated and explicitly cleared by the maintainer
Google Photos AI edit (iOS)
google-photos-ios-edit.jpg is an original 3024 x 4032 iPhone 15 Pro Max photo
edited in Google Photos on iOS to remove an object, supplied by the maintainer
on 2026-09-23 and cleared for publication unmodified, including its EXIF GPS
position. The C2PA manifest is signed by Google LLC with certificate common name
Google Photos (claim generator Google C2PA SDK for iOS) and records
c2pa.opened and c2pa.deleted, the latter typed
compositeWithTrainedAlgorithmicMedia, with no action description and no
SynthID action. XMP adds photoshop:Credit Edited with Google AI. Nothing in
this file names the edit tool or SynthID; Google's checker found SynthID in all
four Photos AI edits under data/captures/2026-09/google-photos/, so the tool
reports it present.
- SHA-256:
d32a9ddb07f6902262246ec380b04e1e7016ef40bbffe6e8d82f4dc48e84ba23 - Expected signal: C2PA AI-enhanced, platform
Google Photos (AI edit), SynthID present with the Google Photos measurement caveat - License: repository license; captured and explicitly cleared by the maintainer
Apple Image Playground
apple-image-playground.png is an original 1024 x 1024 Display P3 PNG from
Image Playground, exported by the maintainer from their iPhone on 2026-09-23.
The file records no OS version; the device ran iOS 27.0 (operator's
Settings screen, 2026-09-23). Apple writes no C2PA manifest and no EXIF software field; the only
provenance is XMP photoshop:Credit Apple Image Playground beside
Iptc4xmpExt:DigitalSourceType trainedAlgorithmicMedia. There is no visible
mark. Apple announced SynthID for Image Playground in a later software update.
Only Apple's own verifier could show whether this file carries it; another
vendor's SynthID oracle is not evidence either way.
- SHA-256:
5ade61fc4cd869e2487745af61c7e23d008a3e9c2b5c440d2ffe16c2d3a32315 - Expected signal: IPTC digitalSourceType, platform
Apple Image Playground - License: repository license; generated and explicitly cleared by the maintainer