EU AI Act Transparency
1. Why this page
Ultido builds digital guest experiences that include AI features – in particular the optional stylisation of guest photos into themed images. The EU AI Act (Regulation (EU) 2024/1689) requires transparency for AI-generated content (Art. 50, applicable since 2 August 2026). This page explains how we implement it.
2. What AI we use
- AI image stylisation (product): on request, a guest photo is turned into a stylised themed image. The computation runs on the external AI service Google Vertex AI. Ultido operates no models of its own for this.
- Currently not in use: the feature is not enabled in any of our deployments. It is not part of the standard scope and is switched on separately per park. Until then, no AI processing of guest photos takes place. The following sections describe what applies from activation onwards.
- No AI on this website: ultido.com itself uses no AI systems towards visitors (no chatbot, no automated decision-making).
3. How we implement Art. 50 of the AI Act
From activation of the feature:
- Labelling: every AI-generated image is labelled visibly as AI-generated. The machine-readable marking is provided by the model provider (Art. 50(2)).
- Information before use: guests give explicit consent and are informed in advance what happens to their photo (platform AI info page).
- No deception: the feature produces clearly stylised themed images – no deepfakes of real situations.
- Before switch-on we complete a data protection impact assessment and evidence the labelling.
4. What we do not do
No facial recognition or biometric identification, no emotion recognition, no social scoring, no prohibited practices within the meaning of Art. 5 AI Act. In our assessment we operate no high-risk AI system (Annex III AI Act). Guest photos are processed only transiently and are not used to train models.
5. Roles and responsibility
Ultido integrates a third-party AI system (Google Vertex AI) into its product and is responsible for its use towards users. Ultido is therefore a deployer within the meaning of the AI Act, not a provider; the provider is Google. This gives rise to the disclosure duty under Art. 50(4). The machine-readable marking under Art. 50(2) is the provider's obligation. The disclosure duty under Art. 50(4) applies to image content that constitutes a deepfake; purely fantastical game content falls outside that definition according to the Commission's guidelines. We label throughout regardless, because the line would otherwise have to be drawn per style class. Human oversight: style catalogues and output are editorially curated; abuse reports: max@ultido.com.
6. Data protection
The data protection side (consent, transience, US transfer on the Vertex route) is governed by the platform Privacy Notice and the AI info page.
7. Status and maintenance
This page is updated whenever our use of AI changes (version line and change history at the end of the page).
Change history
Every version of this text is archived together with its checksum. All versions
| Version | Date | Change |
|---|---|---|
| 3.2.0 | 25.09.2026 | English version fully aligned with the German one; § 5 on roles completed. |
| 3.1.2 | 23.09.2026 | Links in the English version updated to their final addresses. |
| 3.1.1 | 23.09.2026 | Links updated to the new addresses of the legal texts. |
| 3.1.0 | 23.09.2026 | Reasoning in § 5 of the German version written out in full. |
| 3.0.0 | 22.09.2026 | Clarified that the AI feature is currently not active in any park; the text is now a commitment for the case of activation. References linked. |
| 2.1.0 | 16.09.2026 | Role as deployer described in more detail and distinguished from the provider's obligations (Art. 50(2)). |
| 2.0.0 | 16.09.2026 | In-house AI models removed – we use Google Vertex AI only. Our role as deployer under Art. 50(4) AI Act named. |
| 1.0.0 | 10.08.2026 | Earliest archived version. |
Version 3.2.0 · Last updated 25.09.2026