Edit

participant: reportSyntheticMedia

Namespace: microsoft.graph

Important

APIs under the /beta version in Microsoft Graph are subject to change. Use of these APIs in production applications is not supported. To determine whether an API is available in v1.0, use the Version selector.

Report synthetic media detections for a participant in a meeting call. This action is intended for certified third-party audio and video synthetic media detection bots admitted to the meeting call. The detection bot calls this action to flag a participant whose stream it identifies as AI-generated or synthetic media. The detection result is propagated to all participants via roster updates.

After a successful request, the service stores the detection on the participant as a syntheticMediaDetectionInfo object in the participant's syntheticMediaDetection property and delivers it to all participants in a roster update notification. Each request with a new id creates a separate detection record; the service doesn't deduplicate or merge reports. When a participant has multiple detection records, the syntheticMediaDetection property reflects the most recent report.

Third-party bots can invoke this action only when the meeting tenant administrator grants the app the Calls.ReportSyntheticMedia.All application permission.

The detection bot must be admitted to the call before it can call this action. For more information about registering a calling bot and joining calls, see Calls and online meetings. The bot obtains the call ID and participant ID from the call roster and subsequent participant roster update notifications, and uses the call's id as {call-id} and the participant's id as {participant-id} in the request URL.

This API is available in the following national cloud deployments.

Global service US Government L4 US Government L5 (DOD) China operated by 21Vianet
✅ ❌ ❌ ❌

Permissions

Choose the permission or permissions marked as least privileged for this API. Use a higher privileged permission or permissions only if your app requires it. For details about delegated and application permissions, see Permission types. To learn more about these permissions, see the permissions reference.

Permission type Least privileged permissions Higher privileged permissions
Delegated (work or school account) Not supported. Not supported.
Delegated (personal Microsoft account) Not supported. Not supported.
Application Calls.ReportSyntheticMedia.All Not available.

HTTP request

POST /communications/calls/{call-id}/participants/{participant-id}/reportSyntheticMedia
POST /app/calls/{call-id}/participants/{participant-id}/reportSyntheticMedia

Request headers

Name Description
Authorization Bearer {token}. Required. Learn more about authentication and authorization.
Content-Type application/json. Required.

Request body

In the request body, supply a JSON representation of the parameters.

The following table lists the parameters that you can include when you call this action.

Parameter Type Description
contentMetadata mediaMetadata Metadata about the content that was analyzed. Required.
detectionDateTime DateTimeOffset Date and time when the bot performed the detection. This value reflects when the analysis occurred, not when the analyzed media was captured. For media that's recorded and analyzed later, specify the time of analysis. The timestamp type represents date and time information using ISO 8601 format and is always in UTC. For example, midnight UTC on Jan 1, 2024, is 2024-01-01T00:00:00Z. Required.
detections syntheticMediaDetectionDetail collection Per-model or per-segment detection results that support the overall finding. Required.
id Guid Unique identifier for the detection event, in GUID format. The detection bot generates this value, which must be globally unique across reports. Each request that specifies a new id creates a separate detection record. Required.
isMalicious Boolean Indicates whether the detected synthetic media has malicious intent. Set to true when the detection bot considers the AI-generated content to have malicious intent; otherwise, false. Required.
overallConfidence Double Aggregated confidence score that the participant's media is synthetic. The value ranges from 0.0 (low confidence) to 1.0 (high confidence). Each confidence value in the detections collection represents the score for an individual model or segment, whereas overallConfidence represents the aggregated score for the participant. Required.
severity detectionSeverity Qualitative severity banding of the detection. Because severity reflects the same underlying signal as overallConfidence, the two values should generally correspond. Optional. Nullable. Omit the value or set it to null when the severity can't be determined. The possible values are: low, medium, high, unknownFutureValue.

Response

If successful, this action returns a 200 OK response code.

Error responses

This action can return the following common errors. For more information, see Errors in Microsoft Graph.

Status code Error condition
400 The request is malformed, such as an invalid participant ID format or a missing required parameter.
403 The app doesn't have the Calls.ReportSyntheticMedia.All permission, or the detection bot isn't admitted to the call.
404 The specified call or participant isn't found.
429 The app exceeded the request rate limit and is throttled.

A live detection bot can call this action frequently for the same participant. To avoid throttling, report only meaningful changes in the detection rather than every analyzed frame, and limit the reporting rate to no more than one request per target participant per minute.

Examples

Example 1: Report audio synthetic media detection

The following example shows how to report a synthetic media detection for audio content.

Request

The following example shows a request.

POST https://graph.microsoft.com/beta/communications/calls/481f3600-983e-4276-9b59-c1b30ec8d125/participants/550fae72-d251-43ec-868c-373732c2704f/reportSyntheticMedia
Content-Type: application/json

{
  "id": "a1b2c3d4-e5f6-7890-abcd-ef1234567890",
  "detectionDateTime": "2026-03-13T21:00:00Z",
  "severity": "high",
  "overallConfidence": 0.92,
  "isMalicious": true,
  "contentMetadata": {
    "modality": "audio",
    "isRealTime": true,
    "mimeType": "audio/pcm",
    "byteSize": 0,
    "duration": 15,
    "audioMetadata": {
      "sampleRateHz": 16000,
      "bitDepth": 16,
      "channels": 1
    },
    "streamingMetadata": {
      "latencyMs": 50,
      "frameDropRate": 0.0,
      "networkJitterMs": 10
    }
  },
  "detections": [
    {
      "modelName": "DeepfakeDetector-v2",
      "modality": "audio",
      "modelTasks": ["voiceClone"],
      "segment": {
        "startTimeSec": 0,
        "endTimeSec": 15,
        "frameIndices": []
      },
      "confidence": 0.95
    }
  ]
}

Response

The following example shows the response.

HTTP/1.1 200 OK

Example 2: Report video synthetic media detection

The following example shows how to report a synthetic media detection for video content.

Request

The following example shows a request.

POST https://graph.microsoft.com/beta/communications/calls/481f3600-983e-4276-9b59-c1b30ec8d125/participants/550fae72-d251-43ec-868c-373732c2704f/reportSyntheticMedia
Content-Type: application/json

{
  "id": "c3d4e5f6-a7b8-9012-cdef-345678901234",
  "detectionDateTime": "2026-03-13T21:05:00Z",
  "severity": "high",
  "overallConfidence": 0.91,
  "isMalicious": true,
  "contentMetadata": {
    "modality": "video",
    "isRealTime": false,
    "mimeType": "video/mp4",
    "byteSize": 2048000,
    "duration": 30,
    "videoMetadata": {
      "codec": "H.264",
      "frameRate": 30.0,
      "bitrateKbps": 512
    }
  },
  "detections": [
    {
      "modelName": "FaceSwapDetector-v3",
      "modality": "video",
      "modelTasks": ["faceSwap", "lipSync"],
      "segment": {
        "startTimeSec": 0,
        "endTimeSec": 0,
        "frameIndices": [150, 180, 210, 240]
      },
      "confidence": 0.91
    }
  ]
}

Response

The following example shows the response.

HTTP/1.1 200 OK

Example 3: Report multimodal synthetic media detection

The following example shows how to report a real-time detection that combines audio and video analysis. The modality is multimodal, the contentMetadata includes both audioMetadata and videoMetadata (plus streamingMetadata for the live stream), and the detections collection contains one entry per modality.

Request

The following example shows a request.

POST https://graph.microsoft.com/beta/communications/calls/481f3600-983e-4276-9b59-c1b30ec8d125/participants/550fae72-d251-43ec-868c-373732c2704f/reportSyntheticMedia
Content-Type: application/json

{
  "id": "b2c3d4e5-f6a7-8901-bcde-f23456789012",
  "detectionDateTime": "2026-03-13T21:10:00Z",
  "severity": "high",
  "overallConfidence": 0.94,
  "isMalicious": true,
  "contentMetadata": {
    "modality": "multimodal",
    "isRealTime": true,
    "mimeType": "video/mp4",
    "byteSize": 0,
    "duration": 20,
    "audioMetadata": {
      "sampleRateHz": 16000,
      "bitDepth": 16,
      "channels": 1
    },
    "videoMetadata": {
      "codec": "H.264",
      "frameRate": 30.0,
      "bitrateKbps": 512
    },
    "streamingMetadata": {
      "latencyMs": 50,
      "frameDropRate": 0.0,
      "networkJitterMs": 10
    }
  },
  "detections": [
    {
      "modelName": "VoiceCloneDetector-v1",
      "modality": "audio",
      "modelTasks": ["voiceClone"],
      "segment": {
        "startTimeSec": 0,
        "endTimeSec": 20,
        "frameIndices": []
      },
      "confidence": 0.93
    },
    {
      "modelName": "FaceSwapDetector-v3",
      "modality": "video",
      "modelTasks": ["faceSwap", "lipSync"],
      "segment": {
        "startTimeSec": 5,
        "endTimeSec": 18,
        "frameIndices": [150, 180, 210, 240]
      },
      "confidence": 0.95
    }
  ]
}

Response

The following example shows the response.

HTTP/1.1 200 OK
  • Last updated on