Skip to main content
PHOTOTOLOGY

Legal

AI Disclosures

This is a plain-language summary of our AI system disclosures. It is a companion to Section 3A of our Terms of Service, which is the authoritative text. Where this page and the Terms differ, the Terms control.

Phototology is an AI system under Regulation (EU) 2024/1689 (the "EU AI Act"). Article 50 transparency rules start applying on 2 August 2026. We are publishing these disclosures early so you know what you are using.

What this covers

Every analysis you run on phototology.com, analyze.phototology.com, or api.phototology.com is produced by a third-party AI model working on the photo you upload. Results are probabilistic, not determinative. Treat them as a signal, not a verdict.

When you open a lens that triggers an Article 50 disclosure, we show an in-product notice before analysis runs. The banner on analyze pages is part of that notice.

Our AI sub-processors

Your photo is processed by one or more of these providers:

  • Google Gemini
  • OpenAI (GPT-4o and successors)
  • Anthropic Claude

We route each lens to the provider best suited for it. The specific model used for a given result is shown on the result card. Our Privacy Policy lists the sub-processors we use and what they receive.

Markers we write to your images

Natural-language outputs from our lenses (descriptions, captions, atmosphere summaries) are AI-generated. To keep this detectable downstream, we:

  • Write machine-readable AiGenerated markers into image metadata (EXIF, IPTC, and XMP) identifying the source model and lens
  • Return an ai_generated: true field on every lens response from the API
  • Attach C2PA content credentials to enriched images where the output format supports them

We do not change the visible pixels of your photo. Metadata and optional C2PA manifests are the only modifications we make to the file. If a downstream editor strips metadata, the markers may be removed. We recommend preserving metadata when you re-export.

What we don't do

We want to be direct about capabilities we do not offer:

  • No biometric categorisation. The People lens reports presence, count, and general demographics. It does not extract biometric identifiers, build biometric templates, or recognise individuals.
  • No emotion recognition. The Atmosphere lens scores the overall mood and aesthetic of a photograph. It does not infer the emotions or intentions of identifiable natural persons from their biometric data.
  • No deepfakes. We do not generate or manipulate visible image pixels, and we do not produce deepfakes under Article 3(60) of the EU AI Act. If we ever ship a feature that does, we will update this page and apply the Article 50(4) disclosure at the point of generation.

Using the Service to attempt biometric categorisation, emotion recognition, or identification of individuals is prohibited by Section 7 of the Terms. If you do it anyway, you are acting as a deployer of a separately regulated AI system and you take on the Article 50 obligations yourself.

Your rights under the EU AI Act

If you are a user in the European Union, or if your photo depicts people in the EU:

  • You have the right to be told you are interacting with an AI system. The banner on analyze pages and this page are part of that disclosure.
  • You have the right to know that natural-language outputs are artificially generated. The metadata markers and the ai_generated: true API field exist for this reason.
  • These disclosures do not replace your rights under the GDPR or national implementing law. They are additional, not substitute. See Article 50(6) of the EU AI Act.

If you build a product on our API and expose AI-generated output to your own end users, you are a deployer under the EU AI Act. You are independently responsible for Article 50 disclosures to those users, and you must preserve our AI-generated markers in your downstream processing.

Contact

Questions or requests related to AI disclosures: [email protected].

For the full legal text, see Section 3A of the Terms of Service.