VISIONModeration for dating apps, from the first photo to the last message.
Profiles and chats are where a dating app earns trust or loses it. Detectivision checks photos, bios and messages for the flags your rules forbid, and your moderators act on them from one place.
Each place, its flags
| Where | Type | Flags |
|---|---|---|
| Profile photos | image_moderation | sexual, violence, hate, scam |
| Bios and prompts | text_moderation | sexual, hate, racism, sexism, insult |
| Chat messages | text_moderation | sexual, insult, hate, self_harm |
| Photos sent in chat | image_moderation | sexual, scam |
Every account has these flags. GET /api/v1/flags lists the ones yours may send.
At sign-up
Check each profile photo before it is shown. detect mode gives the verdict alone, the cheapest check.
In chat
Send messages with "async": true and the ids of both people in metadata. Flagged ones come back as moderation.flagged webhooks, at half the price.
On a report
Moderators open the request in Logs and take one of your actions, such as Warn or Ban. Your system gets it by webhook and carries it out.
A chat message, checked in the background
curl https://api.detectivision.ai/api/v1/moderation \
-H "Api-Key: $DETECTIVISION_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"type": "text_moderation",
"input": "Send me your photos or I will tell everyone about you.",
"flags": ["sexual", "insult", "hate", "self_harm"],
"async": true,
"metadata": { "senderId": "u_481", "matchId": "m_9920" }
}'{
"id": 6044,
"type": "text_moderation",
"status": "queued",
"metadata": { "senderId": "u_481", "matchId": "m_9920" }
}The API answers at once with the request’s id. The result follows as a moderation.completed webhook, and as moderation.flagged when a flag was raised.
NSFW detection
Nudity and sexual content in photos and messages.
Toxicity detection
Insults, hate and harassment in chat.
Self-harm detection
When a match needs help rather than a ban.
Frequently asked
Can it catch scams?
In images: the flag scam finds pictures made to trick people, such as a fake payment screen or a fake giveaway. Text has no scam flag.
Do you keep the photos?
No. An image is downloaded, checked and not kept.
Which languages are supported?
Turkish and English. Text in other languages is checked, but the results are not reliable.
How do moderators work with it?
Every request shows in the dashboard’s Logs. You define your own actions, such as Approve or Ban; when a moderator takes one, your system gets a moderation.action webhook and carries it out. Detectivision never acts on your platform itself.
Is there a free trial?
Yes. The plans, and the free trial, are on the pricing page.