Cross-service confused deputy prevention

The confused deputy problem is a security issue where an entity that doesn't have permission to perform an action can coerce a more-privileged entity to perform the action. In AWS, cross-service impersonation can result in the confused deputy problem. Cross-service impersonation can occur when one service (the calling service) calls another service (the called service). The calling service can be manipulated to use its permissions to act on another customer's resources in a way it should not otherwise have permission to access. To prevent this, AWS provides tools that help you protect your data for all services with service principals that have been given access to resources in your account.

We recommend using the aws:SourceArn and aws:SourceAccount global condition context keys in resource policies to limit the permissions that Amazon Machine Learning gives another service to the resource. If the aws:SourceArn value does not contain the account ID, such as an Amazon S3 bucket ARN, you must use both global condition context keys to limit permissions. If you use both global condition context keys and the aws:SourceArn value contains the account ID, the aws:SourceAccount value and the account in the aws:SourceArn value must use the same account ID when used in the same policy statement. Use aws:SourceArn if you want only one resource to be associated with the cross-service access. Use aws:SourceAccount if you want to allow any resource in that account to be associated with the cross-service use.

The most effective way to protect against the confused deputy problem is to use the aws:SourceArn global condition context key with the full ARN of the resource. If you don't know the full ARN of the resource or if you are specifying multiple resources, use the aws:SourceArn global context condition key with wildcards (*) for the unknown portions of the ARN. For example, arn:aws:servicename:*:123456789012:*.

The following example shows how you can use the aws:SourceArn and aws:SourceAccount global condition context keys in Amazon ML to prevent the confused deputy problem when reading data from an Amazon S3 bucket.

{
    "Version": "2008-10-17",
    "Statement": [
    {
        "Effect": "Allow",
        "Principal": { "Service": "machinelearning.amazonaws.com" },
        "Action": "s3:GetObject",
        "Resource": "arn:aws:s3:::examplebucket/exampleprefix/*"
        "Condition": {
            "StringEquals": { "aws:SourceAccount": "123456789012" }
            "ArnLike": { "aws:SourceArn": "arn:aws:machinelearning:us-east-1:123456789012:*" } 
        }
    },
    {
        "Effect": "Allow",
        "Principal": {"Service": "machinelearning.amazonaws.com"},
        "Action": "s3:ListBucket",
        "Resource": "arn:aws:s3:::examplebucket",
        "Condition": { 
            "StringLike": { "s3:prefix": "exampleprefix/*" }
            "StringEquals": { "aws:SourceAccount": "123456789012" }
            "ArnLike": { "aws:SourceArn": "arn:aws:machinelearning:us-east-1:123456789012:*" }
        }
    }]
}