AWS sagemaker documentation change

Service: sagemaker · 2025-10-16 · Documentation medium

File: sagemaker/latest/dg/mkt-algo-model-internet-free.md

Summary

Clarified network isolation behavior and S3 access mechanisms when using SageMaker AI execution role

Security assessment

The changes enhance documentation about existing security controls (network isolation and credential management) but do not address a specific security vulnerability. The update provides clearer explanation of how secure S3 access is maintained separately from container operations, which helps users understand security boundaries.

Diff

diff --git a/sagemaker/latest/dg/mkt-algo-model-internet-free.md b/sagemaker/latest/dg/mkt-algo-model-internet-free.md
index be8dce64f..19d2ff384 100644
--- a//sagemaker/latest/dg/mkt-algo-model-internet-free.md
+++ b//sagemaker/latest/dg/mkt-algo-model-internet-free.md
@@ -21 +21,3 @@ Network isolation is required to run training jobs and models using resources fr
-If you enable network isolation, the containers can't make any outbound network calls, even to other AWS services such as Amazon S3. Additionally, no AWS credentials are made available to the container runtime environment. In the case of a training job with multiple instances, network inbound and outbound traffic is limited to the peers of each training container. SageMaker AI still performs download and upload operations against Amazon S3 using your SageMaker AI execution role in isolation from the training or inference container. 
+When you enable network isolation, your training and inference containers can't make any outbound network calls to any service, including Amazon S3. No AWS credentials are made available to the container runtime environment. For training jobs with multiple instances, network inbound and outbound traffic is limited to communication between training container peers.
+
+SageMaker AI still handles all necessary Amazon S3 download and upload operations using your SageMaker AI execution role. This happens apart from your training and inference containers, ensuring that your training data and model artifacts are still accessible while maintaining container isolation.