Enhancing Data Governance: Amazon SageMaker Catalog Introduces Restricted Terms for Governed Classification
Seattle, WA – September 3, 2025 – Amazon Web Services (AWS) today announced a significant enhancement to Amazon SageMaker Catalog, its comprehensive metadata management service for machine learning. The new feature introduces support for governed classification with restricted terms, a powerful addition designed to further strengthen data governance and compliance within machine learning workflows.
This update empowers organizations to establish more granular control over sensitive data and models by defining and enforcing a set of approved or disallowed terms within their SageMaker Catalog classifications. This capability is particularly crucial for industries with stringent regulatory requirements, such as finance, healthcare, and government, where ensuring data accuracy, privacy, and compliance is paramount.
What are Restricted Terms and How Do They Benefit SageMaker Catalog?
Restricted terms provide a mechanism to create curated lists of specific keywords or phrases that can either be mandatorily included or strictly excluded when classifying data assets and models within SageMaker Catalog. This allows for a more sophisticated and controlled approach to metadata management.
Here’s how this new feature can benefit your organization:
- Enhanced Data Privacy and Security: By defining restricted terms, organizations can proactively prevent the accidental or intentional inclusion of sensitive information, such as Personally Identifiable Information (PII) or protected health information (PHI), in publicly accessible or less-controlled catalog entries. This significantly reduces the risk of data breaches and unauthorized access.
- Improved Compliance and Auditing: The ability to enforce specific terminology ensures that data and models are classified according to industry standards and regulatory mandates. This simplifies auditing processes and provides a clear, auditable trail of how data has been classified, aiding in compliance efforts.
- Standardized Metadata and Knowledge Sharing: Restricted terms can be used to enforce the use of standardized terminology for common data attributes, model types, or project categories. This promotes consistency across the organization, making it easier for data scientists, analysts, and other stakeholders to understand, discover, and collaborate on data assets and models.
- Streamlined Model Governance: For machine learning models, restricted terms can be applied to areas like model explainability, bias mitigation efforts, or ethical considerations. This helps ensure that critical governance aspects are consistently documented and communicated.
- Proactive Risk Management: By identifying and restricting potentially problematic terms or concepts, organizations can proactively manage risks associated with data usage and model deployment, fostering a more responsible AI ecosystem.
How it Works:
With this new functionality, users can now define lists of restricted terms within SageMaker Catalog. These lists can be configured to either:
- Mandate Inclusion: Certain essential terms must be present in a classification for it to be considered valid.
- Prohibit Exclusion: Specific terms must not appear in a classification, ensuring sensitive or undesirable information is omitted.
When users attempt to classify data assets or models, SageMaker Catalog will automatically check against these defined restricted term lists, providing alerts or preventing the classification if the rules are violated. This proactive enforcement ensures that metadata remains compliant and accurate from the outset.
A Step Forward for Responsible AI:
Amazon SageMaker Catalog’s introduction of governed classification with restricted terms represents a significant step forward in AWS’s commitment to providing robust tools for responsible and compliant AI development. By offering enhanced control over data classification, AWS empowers organizations to build and deploy machine learning solutions with greater confidence, knowing that their data governance practices are strengthened at every stage.
This feature is now available in Amazon SageMaker Catalog, providing a valuable new capability for organizations looking to elevate their data governance and compliance strategies within their machine learning initiatives.
Amazon SageMaker Catalog adds support for governed classification with restricted terms
AI has delivered the news.
The answer to the following question is obtained from Google Gemini.
Amazon published ‘Amazon SageMaker Catalog adds support for governed classification with restricted terms’ at 2025-09-03 07:00. Please write a detailed article about this news in a polite tone with relevant information. Please reply in English with the article only.