AI transparency
Where Love Jou uses artificial intelligence, where it does not — and what that means under the EU AI Act.
In short
The images and videos on this website were generated with AI. The people in them do not exist. We disclose this visibly on every image and additionally in machine-readable form inside the file itself.
There is exactly one AI feature in the product: the conversational assistant under "Talk". It is optional, off by default, and requires its own explicit consent. Everything else in Love Jou runs without AI.
The assistant sees only what you write to it in that conversation. Your daily questions, answers, moments, chapters and check-ins are not shown to it. Neither are the other person's texts.
None of it is used to train models. There is no scoring of your relationship, no profiling and no automated decision-making about people.
What was generated with AI
The video at the top of the home page (Seedance 2.5), the three short clips in the product sections (Seedance 1 Pro), the cover images of the knowledge articles and the social preview image (Nano Banana Pro).
Every person shown is synthetic. They are neither real couples nor users of Love Jou, and the scenes did not take place.
Alongside the visible label, each file carries the IPTC standard value "trainedAlgorithmicMedia" in the DigitalSourceType field, so search engines, social networks and verification tools can recognise the origin automatically.
Where no AI is involved
The questions do not come from a language model. They were written editorially on the basis of published relationship research and sit unchanged in the source code — readable on the "All questions" page.
Your answers, moments, chapters and check-ins are only stored and shown to the two of you. They are not sent to a model, not summarised, not analysed and not used for training — not even when you use the conversational assistant. It receives only what you write to it directly.
There is no emotion recognition within the meaning of the AI Act: the assistant asks you how you are — it does not infer it from biometric data. There is no personality analysis, no scoring of your relationship and no automated recommendations derived from your stored content. The question of the day is picked by a fixed, readable rule — depth by how many questions came before, dimension by whichever has gone longest unserved — not by a model.
Classification under the AI Act
With the conversational assistant we operate an AI system within the meaning of Art. 3(1) of Regulation (EU) 2024/1689. We are a deployer within the meaning of Art. 3(4) — both for the assistant and for the published synthetic images and video.
Art. 50(1) requires that it be evident you are talking to an AI system. That is stated before the first conversation, under every input field, and the assistant says so itself when asked. Art. 50(4) continues to apply to the synthetic media.
No emotion recognition within the meaning of Art. 3(39): that definition requires inferring emotions from biometric data. The assistant asks questions and reads the answer — it evaluates neither voice nor face nor body data.
None of the practices in Art. 5 apply: no social scoring, no emotion recognition in the workplace or in education, no subliminal manipulation, no biometric categorisation, no remote biometric identification.
None of the high-risk categories in Annex III are touched. The classification is therefore minimal risk, with an active Art. 50 transparency obligation since 2 August 2026.
The assistant makes no decisions about people within the meaning of Art. 22 GDPR. It advises; what follows is yours to decide. It also passes nothing to the other person on its own — a handover happens only when you trigger it.
AI in development
Parts of this website were developed with the help of AI tools. That concerns how it was built, not how it runs: no AI processes your data in the running service.
The environment configuration contains credentials for AI providers. They are used solely to generate the media listed above and are called by no feature of the product.
Questions about this
Write to support@lovejou.com. If you want to know whether a particular image is AI-generated: yes — every photograph of people on this website is.