A company runs automated processes that create SageMaker training jobs. New compliance rules forbid collecting aggregated metadata from training jobs. Which solution will stop SageMaker from collecting metadata for submitted training jobs?

The correct solution is to opt out of metadata tracking for any training job that is submitted. SageMaker allows you to disable metadata collection for individual training jobs by setting the EnableInteractions parameter to False in the ExperimentConfig when creating a training job. This directly addresses the compliance requirement to stop collecting aggregated metadata. Running training jobs in a private subnet in a custom VPC enhances network isolation but does not prevent SageMaker from collecting metadata about the job itself. Encrypting training data with AWS KMS protects the data at rest and in transit but does not stop SageMaker's internal metadata collection processes. Reconfiguring training jobs to use AWS Nitro instances provides enhanced security and performance at the instance level, but it does not control SageMaker's metadata tracking features.