EU AI Act Disclosure
1. Legal framework
WishPride is subject to Regulation (EU) 2024/1689 of 13 June 2024 laying down harmonised rules on artificial intelligence ('EU AI Act'), which entered into force on 1 August 2024. The transparency obligations laid down in article 50, which apply to our activity, become fully binding on 2 August 2026. This page documents our compliance with these obligations and anticipates future developments.
2. Our role under the regulation
WishPride acts as a 'deployer' (art. 3(4) AI Act) of third-party AI systems: we use AI models provided by third parties (Amazon Web Services, Microsoft) without developing or marketing them. We are not a 'provider' (art. 3(3)) of AI systems, nor an 'importer', nor a 'distributor'. The obligations weighing on providers (conformity assessment, CE marking, EU register) therefore do not apply to us.
3. Classification of our AI systems
In accordance with the AI Act's risk-based approach, we classify our systems as follows: • AWS Rekognition (image moderation) — minimal risk under art. 50 (system intended for content moderation, outside annex III). Obligation: user information on use. • Microsoft PhotoDNA (CSAM detection) — minimal risk (child safety tool, outside annex III). Obligation: user information. • Feed ranking algorithm (content recommendation) — minimal risk under art. 50 (recommendation system outside annex III). We do not use any high-risk AI system within the meaning of annex III of the regulation, any prohibited system within the meaning of article 5, nor any in-house general-purpose AI system (GPAI).
4. AI systems used and their purposes
The AI systems deployed on WishPride and their purposes are as follows: • Image moderation (AWS Rekognition): automatic classification of uploaded images into categories (nudity, suggestive content, violence, drugs, hate symbols). Applied before publication. Consequence of a positive detection: refusal of publication or queuing for human validation. • CSAM detection (Microsoft PhotoDNA): perceptual hash-matching of 100% of uploaded images against the NCMEC database. Consequence of a match: automatic blocking + reporting to authorities (see /legal/child-safety). • Feed personalization: ranking algorithm weighted on 8 signals (interactions, follows, hashtags, qualified views). You can view and erase these signals at any time from /settings/privacy/feed-insights. The detail of GDPR legal bases for each of these processings is documented in our Privacy Policy (/legal/privacy).
5. Transparency obligations (art. 50 AI Act)
Article 50 of the AI Act imposes transparency obligations applicable from 2 August 2026: • Art. 50(1) — interaction with an AI system: WishPride currently operates no chatbot or AI conversational agent in direct interaction with its users. Should such a system be deployed in the future, you will be informed in a clear and visible manner before interaction. • Art. 50(2) — AI-generated content: WishPride offers no automatic content generation function (text, image, video, audio) to its users. Should such a function be added, generated content would be labelled as such (visible label + technical metadata compliant with the C2PA standard or equivalent). • Art. 50(4) — deepfakes: WishPride prohibits the publication of non-consensual deepfakes in its Content Policy (/legal/content-policy) and enforces this rule via automated and human moderation.
6. Prohibited practices we do not use
In accordance with article 5 of the AI Act, certain AI practices are prohibited. WishPride confirms it does not use any of them: • Subliminal behavioural manipulation (art. 5(1)(a)). • Exploitation of vulnerabilities linked to age, disability or social situation (art. 5(1)(b)). • Social scoring by public authorities or on their behalf (art. 5(1)(c)). • Emotion recognition in the workplace or in education (art. 5(1)(f)). • Biometric categorisation to infer race, political opinion, sexual orientation, etc. (art. 5(1)(g)). • Real-time remote biometric identification in public spaces (art. 5(1)(h)).
7. Right to human review
In accordance with article 22 of the GDPR and the spirit of the AI Act, you have the right to request human review of any automated decision affecting you (for example: refusal to publish an image by AWS Rekognition, downranking of content, account suspension). To exercise this right, contact appeals@wishpride.com specifying the contested decision and your arguments. The review is free of charge and carried out by a human under the publisher's supervision, within a target time of 14 working days. See also our DSA page (/legal/dsa) for details of the internal complaint-handling system.
8. Future developments and code of conduct
This page will be updated each time a new AI system is deployed on WishPride, as well as at every significant regulatory development. WishPride commits to voluntarily adhering, as soon as they are published, to sectoral codes of conduct provided for in article 95 of the AI Act for providers and deployers of minimal-risk systems, in particular those that would cover social platforms and content moderation. For any question relating to AI use on WishPride, contact dpo@wishpride.com.
