AI Disclosure

1. What Cladegrove makes

Cladegrove is an AI photo studio. Every image produced by the Service is a synthetic image generated by a third-party AI model. The persona that the model is asked to render can be built either from a textual description (modes text_free and text_guided) or from a reference photo uploaded by the user (mode photo_upload). In either case the output is a fresh AI image, not a pixel-faithful copy of any uploaded photo, but in the photo-upload case the identity of the rendered subject is anchored to a real likeness.

The photo-upload mode is gated by an explicit consent declaration linked to our Likeness Policy: the user confirms that the subject is themselves or a third party who has given explicit consent, that the subject is an adult, and that the user takes full responsibility for the upload and any output derived from it. The uploaded reference is stored privately and deleted when the related character is deleted.

See our Terms of Service §5 for the Acceptable Use rules that prohibit attempts to impersonate, defame, sexualise or otherwise misuse the likeness of real persons, including by way of the photo-upload character flow.

2. How Cladegrove images are marked (Art. 50 EU AI Act)

Pursuant to Article 50(2) of Regulation (EU) 2024/1689 (the EU AI Act), generated outputs must be "marked in a machine-readable format and detectable as artificially generated or manipulated". Every image produced by Cladegrove carries two independent provenance layers, listed below.

2.1 SynthID watermark (in the pixels)

SynthID is an imperceptible watermark embedded directly into the pixels of the image by the upstream model that synthesises it: OpenAI's gpt-image model, using content provenance technology developed in partnership with Google DeepMind. The marking is applied at generation time, by the model, and is not added afterwards by Cladegrove. It does not change how the image looks to the human eye, but it can be read by SynthID-aware detection tooling. Because it lives in the image content itself, it survives common transformations such as re-compression, resizing, cropping and screenshots that would normally strip file metadata.

2.2 IPTC and XMP provenance metadata (in the file)

Before storing an output, Cladegrove writes machine-readable provenance metadata into the file itself, in both EXIF and XMP. This layer is added on our side and is the primary metadata-level marker that travels with the file:

• an IPTC DigitalSourceType field set to trainedAlgorithmicMedia— the controlled-vocabulary value used by the IPTC and referenced in the AI Act guidance to flag AI-generated content;
• a dedicated GenAI XMP namespace recording the upstream model identifier (e.g. gpt-image-2 or gpt-5.5), the generation ID, the creation timestamp, links to our Terms and to this AI Disclosure page, and a synthetic-media disclaimer;
• standard EXIF fields (Software, Artist, Copyright, ImageDescription) consistently declaring the file as AI-generated.

Together the two layers cover the two realistic failure modes: file metadata being stripped, where the SynthID pixel watermark still carries the signal, and pixels being lightly edited, where the IPTC/XMP record still attributes the file to an AI model. For background on the upstream pixel watermark, see OpenAI's note on advancing content provenance.

3. How to verify a Cladegrove image

You can check the AI provenance of any file produced by Cladegrove without installing anything elaborate:

• OpenAI verification tool: upload the image to openai.com/research/verify to check whether it carries OpenAI provenance signals, including the SynthID watermark, even when the file metadata has been removed.
• EXIF/XMP viewers: any metadata viewer reveals the Cladegrove provenance layer. Command-line tools such as exiftool (exiftool image.jpg | grep -i "digital source\|software\|artist") show the IPTC DigitalSourceType = trainedAlgorithmicMedia field, the EXIF Software = Cladegrove and Artist = AI-generated by Cladegrove fields, and the full GenAI XMP namespace with the upstream model, generation ID and timestamps.
• File properties: the standard file-info panels on macOS and Windows surface the EXIF Software, Artist and Copyright fields while the metadata is intact.

4. Caveat: markers can still be lost

Several social platforms, messaging apps, image-compression CDNs and screenshot tools strip file metadata when re-encoding images. Where this happens, the IPTC/XMP layer written by Cladegrove (see §2.2) may be lost. The SynthID pixel watermark is more robust because it lives in the pixels rather than the metadata, but heavy editing, aggressive re-compression or AI-based transformations can still degrade it.

Loss of a marker does not change the AI-generated nature of the content; it changes only the ability of automated tools to detect it. If an image has passed through a third-party platform, treat the verification tools in §3 as best-effort and rely on the visible disclosure described in §5.

5. Obligations of users who publish outputs (Art. 50(4) AI Act)

When a user of Cladegrove publishes, shares or otherwise distributes a Synthetic Output to third parties, the user becomes a deployer under Art. 50(4) of the AI Act. Where the output could be perceived as authentic photography of a real person, situation or event, the deployer is responsible for labelling the content as artificially generated or manipulated in a clear and distinguishable manner, at the latest at the time of the first interaction or exposure.

The provenance signals carried by each output assist, but do not replace, that user-side disclosure obligation. In contexts where file metadata is likely to be stripped (most social platforms, messaging), users should add a visible disclosure: a caption ("AI generated"), an on-image label or a watermark.

See Terms of Service §7 for the full deployer clause.

6. Reporting and right to object

If you believe a Synthetic Output produced by Cladegrove incidentally resembles you to a degree that is recognisable to a third party, or if you wish to report a suspected violation of our Acceptable Use rules (impersonation, NCII, deepfake, content depicting minors, identity replication of a real person), write to privacy@cladegrove.com or to support@cladegrove.com. Provide a description of the relevant generation (and, where available, the URL or file). We will remove the content within 7 days of verification and, in cases of policy violation, terminate the offending account.

7. Updates

This page is updated whenever the technical marking of outputs changes. Marking is performed by the upstream image model, so its exact behaviour may evolve as that model is updated.