AI Notice (draft)

1. What is AI

Everything a matchmaker says, everything the other matchmaker says, the assessments and the introduction reports are written by AI (a large language model). The screen shows them as AI.

The other side of a practice conversation is an invented person, not a real one. It is shown as that.

2. What the AI does

• It speaks about you in the third person, using the public cards you confirmed. It does not pretend to be you.
• It talks with the other matchmaker in fixed rounds (an introduction, a values topic, a lifestyle topic, two situations, a checklist).
• Each matchmaker privately assesses whether the other person suits its own member. For this it may use the private judgment cards only you have; those cards and the assessment are not part of any request made for the other matchmaker.
• If neither assessment says “pass” (each says recommend or maybe), a report is written and sent to each side. If either says “pass”, the two are not paired again for a year.

3. On what basis people are introduced

First, code picks the candidates you could meet from the conditions you set (age range, who you look for, region, languages, lifestyle answers such as smoking, drinking and what you are looking for, tastes you marked as must-match) and from an anonymity floor. People you blocked or who blocked you, people introduced to you or judged not to fit within the last year, and people who joined with the same invite code (unless both of you allow it in your settings and each looks for one opposite gender only) are left out. Then the AI matchmakers talk and each assesses. An introduction goes to the one person who looks like the best fit for both, and a mission has at most two introductions.

The AI's assessment must point to what was said in the conversation, and statements without support are filtered out. A guess is shown as a guess.

The AI never sees members' photos, and photos play no part in who is introduced. A photo is shown only between two introduced members who have both added one, while they decide.

4. Limits

The AI can be wrong or misunderstand. A person does not read every conversation: there is no tool for opening them, and what people read is reports (what the reporter wrote) and counts. Being introduced by an AI does not mean the other person is safe or right for you.

5. What you can do

• If you turned watching on you can replay the conversations your matchmaker had.
• You can decline any introduction without giving a reason.
• You can report a strange or unpleasant conversation or introduction.
• Ask for an explanation: write to the address below to learn why you were (or were not) introduced. We will tell you the criteria used and, as far as we can, which of your information had an effect. [The time within which we answer is to be set.]
• Ask for a person to look again: if you object to an outcome concerning an introduction you can ask a person to review it.

6. Decisions with legal effect

Wingvoy's AI does not decide anything that sets your rights or legal status. What it does is suggest whom to introduce you to. Whether to accept is yours.

One exception: if the safety check by an AI model sees a possible minor or sexual content in a matchmaker's lines or in a card, the member is stopped at once, without any report. That check can be wrong, so we aim for a person to look within 24 hours, and you can object at the address below.

7. Contact

[operator name: not set]

[support email: not set]