AI Feature Terms

Effective: 8 October 2026Version: 1.4

All legal documents
On this page

Artificial intelligence works in several places inside the Service. This document sets out what happens to the text you enter, who owns the output, what must not be entered, and where liability ends. It forms an integral part of the Terms; where the two differ, the more specific rule here governs the AI features.

It is a separate document because AI features change faster than the rest of the contract: this way a model can be swapped or a feature extended without re-accepting the whole Terms.

The In short summaries are informational and are there to help you follow the section. The full text always prevails legally.

This document is available in Hungarian and English. In case of any discrepancy, the Hungarian version prevails, except where the English version is more favourable to a Consumer.

1. Where AI operates in the Service

In shortA chat assistant, suggestions, drafted text, document analysis and spreadsheet column recognition. AI output is labelled as AI.

The Service currently uses a language model supplied by OpenAI (at the date of this version, models of the GPT-6 family: GPT-6.1 Sol and GPT-6 Luna); the model, its version and its provider may change under Section 7. AI appears in the following places, and only these send content for AI processing:

  • conversational assistant (budget, cultural and religious questions);
  • suggestion generation (timeline, tasks, budget, honeymoon, seating, moodboard, and the preview shown after onboarding);
  • textual analysis of vendor quotes and contracts;
  • drafting text you ask for in the chat (for example a message to guests), and the planner consultation questionnaire;
  • column recognition in spreadsheet import, which receives only the headers and up to five masked sample rows (Privacy Policy 2.9).

For transparency under Article 50 of Regulation (EU) 2024/1689 on artificial intelligence and other applicable law, the Service takes two measures: on the conversational surface you are told at the start that you are interacting with an AI system, and AI-generated or AI-assisted content carries a visible label. That label is not decoration: it must not be removed or hidden where it is required to remain visible. The label does not mean that the output has been verified, reviewed or approved by the Provider.

The Service does not generate images, audio or video depicting real people, objects, places or events. If that changes, the affected content will also carry the deepfake marking required by that Regulation, and this document will be updated.

2. Who owns input and output

In shortInput stays yours, and we assign you every transferable right in the output. One thing we cannot promise: that nobody else will ever receive a similar output.

Input is everything you supply to an AI feature: the question, wedding data, uploaded documents. Input remains yours and under your control; the Provider processes it only as far as fulfilling the request requires, within the licence in Section 11 of the Terms.

Output is what the AI produces in response. To the extent the Provider owns or controls any transferable rights in the output, it assigns them to you, and you may use the output for your wedding, your communication and your public wedding pages, subject to these terms, the law and the rights of third parties. Checking whether the output suits your intended use and does not infringe anyone's rights is your responsibility (Section 5).

Two limits have to be stated, because without them the sentence above would promise more than can be kept:

  • By the nature of language models, another user asking a similar question may receive a similar or identical output. Your ownership therefore does not extend to stopping others from using a similar output.
  • Whether purely machine-generated content attracts copyright at all depends on the legal system. The Provider therefore does not warrant that the output is protected by copyright.

3. We do not train on your data

In shortNeither the Provider nor the model provider uses your input or output to train models. This is a warranty, not a statement of intent.

The Provider warrants that it will not use your input or the corresponding output to train, fine-tune or improve any artificial intelligence model, and will not permit the model provider it engages to do so either.

The current model provider, OpenAI, does not use API customer data for model training under its own API policy. It may keep requests for a limited period for abuse and security monitoring, by default up to 30 days, unless longer retention is required by law; the details and the transfer basis are on the Subprocessors and recipients page. The Provider may replace the model provider only if the same undertaking applies to the new one; you are notified of such a change under Section 7.

On our side, AI calls are logged (time, feature, account identifier) solely for abuse prevention, rate limits and cost control, for at most 90 days, as set out in Section 4 of the Privacy Policy.

4. What must not be entered

In shortDo not put health data, government identifiers or payment card numbers into a request, and do not paste sensitive information about other people.

When using the AI features you agree not to enter:

  • special category data under Article 9 GDPR (health, religious or philosophical beliefs tied to a named person, sex life, biometric data) - including information about a guest's health - unless a dedicated field of the Service asks for it and the required explicit consent has been given. The only such case today is the couple's own cultural and religious background, entered in the dedicated field with the separate consent described in Privacy Policy 2.2; the cultural advisor uses it within the scope of that consent. This exception does not cover information about guests, vendors or other third parties;
  • government identifiers (ID card, passport, tax or social security number);
  • payment card or bank account numbers;
  • personal data of third parties beyond what answering the request actually requires;
  • content that infringes anyone's rights, or documents you are not entitled to upload.

About dietary requirements. The Service holds guests' dietary needs as part of guest data, because wedding planning requires it. Do not, however, put the medical reason, diagnosis or condition behind a preference into an AI request. A label such as "gluten-free" is enough for planning; but because even such a label can reveal a health condition, treat a dietary requirement that points to one as sensitive and leave it out of AI requests.

If you breach this, the resulting data protection consequences are yours; the Provider does not pre-screen input, because doing so would mean reading all of it.

5. Accuracy and human review

In shortAI gets things wrong, and does so confidently. Review the output before you act on it or pass it on - and consult a professional for financial, legal or health questions.

The Provider does not warrant the accuracy, completeness, reliability or currency of the output, nor that it does not infringe third-party rights or breach the law. Language models are probabilistic: output may contain false assertions about real places, people or facts, stated confidently.

You agree to apply human review to the output before making a decision based on it, sending it to your guests, or publishing it. This applies with particular force to quote and contract analysis: the analysis is a starting point, not a legal opinion.

AI content does not constitute legal, financial, tax, medical, dietary safety or religious professional advice. Wedding planning decisions are made at your own risk; consult a qualified professional where appropriate.

6. Prohibited use

In shortDo not pass output off as human-written, do not present it as endorsed by us, and do not build automated legally binding decisions on it.

In connection with the AI features it is prohibited to:

  • remove, hide or alter an AI label where the law or the Service requires it to remain visible, or knowingly present AI-generated content as entirely human-written where that would be material to the recipient's decision or understanding (editing and adopting a draft as your own text is not prohibited);
  • present output as verified, approved or recommended by the Provider;
  • build a fully automated decision, without human involvement, that is legally binding or similarly significantly affects the recipient;
  • use the AI features through automated means, in bulk, or for resale (bots, scripts, scraping, embedding into your own product);
  • use them to circumvent model safeguards, attack another system, or produce content prohibited by the Acceptable Use Policy.

7. Limits, model changes, liability

In shortThere are quotas, and the model may change. If a change materially disadvantages you, you can terminate free of charge within 30 days.

Limits. Technical rate limits apply to AI calls for abuse and cost prevention. One AI request is one call to an AI feature (for example a chat message, a generated set of suggestions or a document analysis). The Free plan includes 3 AI requests per rolling 30 days. Pro and Lifetime have no monthly quota (this is what "unlimited" means on the pricing page); a technical limit of 300 requests a day, 1500 a day on the Planner plan, and 30 requests a minute on every plan protects against abuse. A limit is not a promise that every request produces output. Current values are shown on the pricing page.

Model and feature changes. The Provider may replace the language model, its version or its provider, and may modify or discontinue an AI feature; a change may also mean a materially lower AI limit or different data processing arrangements, in which case the change rules of the Privacy Policy and the Subprocessor list apply as well. Where the change disadvantages you to more than a minor extent, the Provider notifies you in advance on a durable medium and you may terminate free of charge within 30 days of the notice. This follows Hungarian Government Decree 373/2021 (VI. 30.) on contracts for the supply of digital content and digital services.

Liability. The liability rules and the limitation of liability in Section 13 of the Terms apply to the AI features, including all of its exceptions. To the extent permitted by law, the Provider is not responsible for professional, planning or other decisions resulting solely from reliance on output that you did not review. This does not exclude or limit liability that cannot lawfully be excluded or limited, including liability for intentional misconduct, gross negligence, death or personal injury, mandatory consumer rights, non-conformity of the Service and data protection obligations.