Core API

AI Content Moderation

Relay integrates Google Cloud Video Intelligence to scan live and recorded sessions for inappropriate content. Confidence-scored flags appear in your dashboard and fire webhooks in real time. Available on the Growth plan and above.

AI moderation requires the Growth plan. Attempting to enable it on Starter returns 403 Forbidden.

Enable moderation

Set moderation: true when creating a hall:

curl

curl -X POST https://api.relay.dev/v1/halls \
  -H "Authorization: Bearer relay_live_..." \
  -H "Content-Type: application/json" \
  -d '{
    "name": "monitored-session",
    "moderation": true
  }'

Relay samples video frames during the live session and again processes the completed recording. Flags are generated as soon as content is analysed — typically within seconds for live flags.

Flag types

Relay uses Google Video Intelligence's EXPLICIT_CONTENT_DETECTION and OBJECT_TRACKING features. Each flag includes a confidence score from 0–100%.

EXPLICIT_CONTENTVERY_LIKELY / LIKELY / POSSIBLE / UNLIKELY

Sexually explicit or adult imagery detected in the video feed.

VIOLENCEconfidence_pct

Violent content or weapons detected.

OFFENSIVE_LANGUAGEconfidence_pct

Detected via audio transcription.

The moderation flag object

flag_idstring

Unique identifier for this flag.

hall_idstring

The hall where this flag was raised.

source"live" | "post_session"

Whether the flag was raised during the live session or during post-processing of the recording.

flag_typestring

The type of content detected (see above).

likelihood_labelstring

Google's likelihood label: VERY_LIKELY, LIKELY, POSSIBLE, UNLIKELY.

confidence_pctnumber

Confidence score from 0 to 100.

frame_timestamp_msinteger

Timestamp within the session where this content was detected, in milliseconds.

status"open" | "dismissed" | "actioned"

Review status. Updated by your team from the dashboard or via API.

reviewed_atdatetime | null

When a reviewer actioned this flag.

Example flag object

{
  "flag_id": "flag_01j9xz...",
  "hall_id": "h_01j9x2...",
  "source": "live",
  "flag_type": "EXPLICIT_CONTENT",
  "likelihood_label": "LIKELY",
  "confidence_pct": 78.4,
  "frame_timestamp_ms": 45200,
  "status": "open",
  "reviewed_at": null,
  "created_at": "2025-04-05T10:30:22.000Z"
}

Billing

Moderation is billed at Google Cloud Video Intelligence cost + 20% margin. Usage is summarised on your monthly invoice with a per-clip breakdown. You can view this month's moderation cost on the Dashboard → Usage page.