AWS managed policies for SageMaker AI training session APIs - Amazon SageMaker AI
AmazonSageMakerTrainingSessionAccessAmazonSageMakerTrainingSessionFullAccessTraining sessions policy updates
Services or capabilities described in AWS documentation might vary by Region. To see the differences applicable to the AWS European Sovereign Cloud Region, see the AWS European Sovereign Cloud User Guide.

AWS managed policies for SageMaker AI training session APIs

Use these AWS managed policies to create and interact with Amazon SageMaker AI training session APIs. With AmazonSageMakerTrainingSessionAccess, you can manage training sessions and run training and sampling operations. With AmazonSageMakerTrainingSessionFullAccess, your training session execution role can access data in Amazon S3, use customer managed AWS Key Management Service (AWS KMS) keys, publish model packages, and manage Amazon VPC network interfaces.

AWS managed policy: AmazonSageMakerTrainingSessionAccess

With this policy, you can create and manage SageMaker AI training sessions and run training and sampling operations against them. Attach this policy to the IAM users and roles that create and interact with training sessions.

Permissions details

This policy includes the following permissions.

  • sagemaker – Allows creating, describing, stopping, deleting, and tagging training sessions. Restricted to resources in the same account.

  • sagemaker – Allows running training and sampling operations against a training session, retrieving their results, and invoking sample requests. Restricted to resources in the same account.

  • sagemaker – Allows listing training sessions. Restricted to resources in the same account.

  • sagemaker – Allows calling SageMaker AI training session runtime APIs with a bearer token. Restricted to resources in the same account.

  • kms – Allows the caller identity to use customer managed AWS KMS keys with training sessions at runtime. The Decrypt and GenerateDataKey actions are restricted through kms:ViaService to SageMaker AI only. Restricted to resources in the same account.

  • iam – Allows passing an execution role to SageMaker AI. The iam:PassRole action is restricted to roles under the service-role path whose names contain SageMaker, is restricted through iam:PassedToService to the SageMaker AI training session service only, and applies only when the role is associated with a SageMaker AI resource. Restricted to resources in the same account.

  • sagemaker – Allows creating model package groups, deleting model packages, describing model packages and model package groups, and listing model packages, for checkpoint publication workflows. Restricted to resources in the same account.

  • sagemaker – Allows describing and listing public hub content, such as pre-built models, in the AWS owned SageMaker AI public hub.

  • s3 – Allows creating the SageMaker AI default Amazon S3 bucket, listing its contents, and deleting output objects in buckets whose names begin with sagemaker-. Restricted to resources in the same account.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "SageMakerTrainingSessionControlPlanePermissions",
            "Effect": "Allow",
            "Action": [
                "sagemaker:CreateTrainingSession",
                "sagemaker:DescribeTrainingSession",
                "sagemaker:StopTrainingSession",
                "sagemaker:DeleteTrainingSession",
                "sagemaker:AddTags"
            ],
            "Resource": "arn:aws:sagemaker:*:*:training-session/*",
            "Condition": {
                "StringEquals": {
                    "aws:ResourceAccount": "${aws:PrincipalAccount}"
                }
            }
        },
        {
            "Sid": "SageMakerTrainingSessionRuntimePermissions",
            "Effect": "Allow",
            "Action": [
                "sagemaker:ExecuteTrainingOperation",
                "sagemaker:GetTrainingOperationResult",
                "sagemaker:ExecuteSamplingOperation",
                "sagemaker:GetSamplingOperationResult",
                "sagemaker:SampleTrainingSession",
                "sagemaker:SampleTrainingSessionWithResponseStream"
            ],
            "Resource": "arn:aws:sagemaker:*:*:training-session/*",
            "Condition": {
                "StringEquals": {
                    "aws:ResourceAccount": "${aws:PrincipalAccount}"
                }
            }
        },
        {
            "Sid": "SageMakerTrainingSessionRuntimeKmsPermissions",
            "Effect": "Allow",
            "Action": [
                "kms:Decrypt",
                "kms:GenerateDataKey"
            ],
            "Resource": "arn:aws:kms:*:*:key/*",
            "Condition": {
                "StringEquals": {
                    "aws:ResourceAccount": "${aws:PrincipalAccount}"
                },
                "StringLike": {
                    "kms:ViaService": "sagemaker.*.amazonaws.com"
                }
            }
        },
        {
            "Sid": "SageMakerTrainingSessionListPermissions",
            "Effect": "Allow",
            "Action": [
                "sagemaker:ListTrainingSessions"
            ],
            "Resource": "*",
            "Condition": {
                "StringEquals": {
                    "aws:ResourceAccount": "${aws:PrincipalAccount}"
                }
            }
        },
        {
            "Sid": "BearerTokenPermissions",
            "Effect": "Allow",
            "Action": [
                "sagemaker:CallWithBearerToken"
            ],
            "Resource": "*",
            "Condition": {
                "StringEquals": {
                    "aws:ResourceAccount": "${aws:PrincipalAccount}"
                }
            }
        },
        {
            "Sid": "PassExecutionRolePermissions",
            "Effect": "Allow",
            "Action": [
                "iam:PassRole"
            ],
            "Resource": [
                "arn:aws:iam::*:role/service-role/*SageMaker*",
                "arn:aws:iam::*:role/service-role/*Sagemaker*",
                "arn:aws:iam::*:role/service-role/*sagemaker*"
            ],
            "Condition": {
                "ArnLike": {
                    "iam:AssociatedResourceArn": "arn:aws:sagemaker:*:*:*"
                },
                "StringEquals": {
                    "aws:ResourceAccount": "${aws:PrincipalAccount}",
                    "iam:PassedToService": [
                        "training-session.sagemaker.amazonaws.com"
                    ]
                }
            }
        },
        {
            "Sid": "SageMakerTrainingSessionModelPackagePermissions",
            "Effect": "Allow",
            "Action": [
                "sagemaker:CreateModelPackageGroup",
                "sagemaker:DeleteModelPackage",
                "sagemaker:DescribeModelPackageGroup",
                "sagemaker:DescribeModelPackage",
                "sagemaker:ListModelPackages"
            ],
            "Resource": [
                "arn:aws:sagemaker:*:*:model-package/*",
                "arn:aws:sagemaker:*:*:model-package-group/*"
            ],
            "Condition": {
                "StringEquals": {
                    "aws:ResourceAccount": "${aws:PrincipalAccount}"
                }
            }
        },
        {
            "Sid": "SageMakerTrainingSessionPublicHubPermissions",
            "Effect": "Allow",
            "Action": [
                "sagemaker:DescribeHubContent",
                "sagemaker:ListHubContents"
            ],
            "Resource": [
                "arn:aws:sagemaker:*:aws:hub/SageMakerPublicHub",
                "arn:aws:sagemaker:*:aws:hub-content/SageMakerPublicHub/*"
            ]
        },
        {
            "Sid": "SageMakerTrainingSessionS3Permissions",
            "Effect": "Allow",
            "Action": [
                "s3:CreateBucket",
                "s3:DeleteObject",
                "s3:ListBucket"
            ],
            "Resource": [
                "arn:aws:s3:::sagemaker-*",
                "arn:aws:s3:::sagemaker-*/*"
            ],
            "Condition": {
                "StringEquals": {
                    "s3:ResourceAccount": "${aws:PrincipalAccount}"
                }
            }
        }
    ]
}
Show moreShow less

For more information, see AmazonSageMakerTrainingSessionAccess in the AWS Managed Policy Reference Guide.

AWS managed policy: AmazonSageMakerTrainingSessionFullAccess

With this policy, your SageMaker AI training session execution roles can access the resources needed to run training sessions. Attach this policy to IAM roles that you use as SageMaker AI training session execution roles.

Permissions details

This policy includes the following permissions.

  • s3 – Allows training sessions to read input data from, write output data to, and list the contents of Amazon S3 buckets. Restricted to resources in the same account.

  • kms – Allows training sessions to decrypt and generate data keys for server-side encryption of Amazon S3 objects using customer managed AWS KMS keys. The Decrypt and GenerateDataKey actions are restricted through kms:ViaService to Amazon S3 only. DescribeKey is allowed without the kms:ViaService restriction. Restricted to resources in the same account.

  • sagemaker – Allows describing public hub content, such as pre-built models, in the AWS owned SageMaker AI public hub.

  • sagemaker – Allows accessing, creating, and describing model packages, and describing model package groups for checkpoint publication workflows. Restricted to resources in the same account.

  • ec2 – Allows managing Amazon VPC network interfaces for training sessions running in Amazon VPC mode, including creating, deleting, and describing network interfaces and related Amazon VPC resources. The ec2:CreateTags action is restricted to the CreateNetworkInterface action. Restricted to resources in the same account.

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "S3Permissions",
            "Effect": "Allow",
            "Action": [
                "s3:GetObject",
                "s3:PutObject",
                "s3:ListBucket"
            ],
            "Resource": "*",
            "Condition": {
                "StringEquals": {
                    "s3:ResourceAccount": "${aws:PrincipalAccount}"
                }
            }
        },
        {
            "Sid": "KMSPermissions",
            "Effect": "Allow",
            "Action": [
                "kms:Decrypt",
                "kms:GenerateDataKey"
            ],
            "Resource": "arn:aws:kms:*:*:key/*",
            "Condition": {
                "StringEquals": {
                    "aws:ResourceAccount": "${aws:PrincipalAccount}"
                },
                "StringLike": {
                    "kms:ViaService": "s3.*.amazonaws.com"
                }
            }
        },
        {
            "Sid": "KMSDescribeKey",
            "Effect": "Allow",
            "Action": "kms:DescribeKey",
            "Resource": "arn:aws:kms:*:*:key/*",
            "Condition": {
                "StringEquals": {
                    "aws:ResourceAccount": "${aws:PrincipalAccount}"
                }
            }
        },
        {
            "Sid": "SageMakerPublicHubReadPermissions",
            "Effect": "Allow",
            "Action": [
                "sagemaker:DescribeHubContent"
            ],
            "Resource": [
                "arn:aws:sagemaker:*:aws:hub/SageMakerPublicHub",
                "arn:aws:sagemaker:*:aws:hub-content/SageMakerPublicHub/*"
            ]
        },
        {
            "Sid": "SageMakerModelPackagePermissions",
            "Effect": "Allow",
            "Action": [
                "sagemaker:AccessModelPackage",
                "sagemaker:CreateModelPackage",
                "sagemaker:DescribeModelPackage",
                "sagemaker:DescribeModelPackageGroup"
            ],
            "Resource": [
                "arn:aws:sagemaker:*:*:model-package/*",
                "arn:aws:sagemaker:*:*:model-package-group/*"
            ],
            "Condition": {
                "StringEquals": {
                    "aws:ResourceAccount": "${aws:PrincipalAccount}"
                }
            }
        },
        {
            "Sid": "EC2NetworkPermissions",
            "Effect": "Allow",
            "Action": [
                "ec2:CreateNetworkInterface",
                "ec2:CreateNetworkInterfacePermission",
                "ec2:DescribeNetworkInterfaces",
                "ec2:DescribeVpcs",
                "ec2:DescribeSubnets",
                "ec2:DescribeSecurityGroups",
                "ec2:DescribeDhcpOptions"
            ],
            "Resource": "*",
            "Condition": {
                "StringEquals": {
                    "aws:ResourceAccount": "${aws:PrincipalAccount}"
                }
            }
        },
        {
            "Sid": "EC2NetworkInterfaceTagPermissions",
            "Effect": "Allow",
            "Action": [
                "ec2:CreateTags"
            ],
            "Resource": "arn:aws:ec2:*:*:network-interface/*",
            "Condition": {
                "StringEquals": {
                    "aws:ResourceAccount": "${aws:PrincipalAccount}",
                    "ec2:CreateAction": "CreateNetworkInterface"
                }
            }
        },
        {
            "Sid": "EC2NetworkInterfaceDeletePermissions",
            "Effect": "Allow",
            "Action": [
                "ec2:DeleteNetworkInterface",
                "ec2:DeleteNetworkInterfacePermission"
            ],
            "Resource": "*",
            "Condition": {
                "StringEquals": {
                    "aws:ResourceAccount": "${aws:PrincipalAccount}"
                }
            }
        }
    ]
}
Show moreShow less

For more information, see AmazonSageMakerTrainingSessionFullAccess in the AWS Managed Policy Reference Guide.

Amazon SageMaker AI updates to SageMaker AI training sessions managed policies

View the following table for a history of AWS managed policies for SageMaker AI training sessions, including each policy version and the date it was added.

Policy Version Change Date

AmazonSageMakerTrainingSessionAccess – New policy

1

Initial policy (version 1).

 October 13, 2026

AmazonSageMakerTrainingSessionFullAccess – New policy

1

Initial policy (version 1).

 October 13, 2026