AI Guardrails:Configure notifications

Webhook address requirements

Your webhook address must meet all of the following requirements:

• Accessible from the Internet over HTTP or HTTPS

• Supports the POST method

• Supports UTF-8 encoding

• Accepts data in application/x-www-form-urlencoded format

• Handles the checksum and content form parameters

Callback payload fields

Each callback POST body contains two form parameters:

Verifying the checksum

On receipt of a callback, recompute the signature using the same algorithm — SHA256(<User UID> + <seed> + <content>) — and compare it against the checksum field. If the values differ, discard the request.

The following Python example shows how to verify the checksum:

import hashlib

def verify_checksum(user_uid: str, seed: str, content: str, received_checksum: str) -> bool:
    """Verify that a callback originated from AI Guardrails."""
    raw = user_uid + seed + content
    computed = hashlib.sha256(raw.encode("utf-8")).hexdigest()
    return computed == received_checksum

Procedure

1. Set up a webhook address that meets all the requirements listed in Webhook address requirements, and choose a seed value.

2. When calling an async AI Guardrails API operation, include the callback parameter (your webhook address) and the seed parameter in the request body. Refer to the individual API operation documentation for parameter details.

Encryption algorithms

scanResult structure

The scanResult structure depends on the type of content that was scanned:

• Images: Same structure as the results parameter returned by synchronous image scans.

• Videos: Same structure as the results parameter returned by asynchronous video scans.

auditResult structure

humanAuditResult structure

Example

The following shows a fully populated content object after parsing:

{
    "scanResult": {
        "code": 200,
        "msg": "OK",
        "taskId": "fdd25f95-4892-4d6b-aca9-7939bc6e9baa-1486198766695",
        "url": "http://1.jpg",
        "results": [
            {
                "rate": 100,
                "scene": "porn",
                "suggestion": "block",
                "label": "porn"
            }
        ]
    },
    "auditResult": {
        "suggestion": "block",
        "labels": [
            "porn",
            "ad",
            "terrorism"
        ]
    },
    "humanAuditResult": {
        "suggestion": "pass",
        "dataId": "yyyy",
        "labels": [
            "porn",
            "vulgar"
        ],
        "taskId": "xxxxxx"
    }
}