Microsoft Tackles AI Data Risks in Fabric
▼ Summary
– New Microsoft Purview innovations for Microsoft Fabric help secure data and accelerate AI adoption by focusing on risk identification, preventing data oversharing, and strengthening governance.
– Integration between Purview and Fabric provides unified security and governance, protecting sensitive data and ensuring data readiness for AI.
– Data Loss Prevention (DLP) policies in Fabric prevent data oversharing by identifying sensitive data and restricting access in workloads like Warehouse and databases.
– Insider Risk Management extends to Fabric lakehouses, using built-in risk indicators to detect potential data exfiltration and other risky user actions.
– Microsoft Purview improves governance and data quality through a Unified Catalog for asset visibility and built-in capabilities to measure, monitor, and enhance data quality.
Microsoft’s latest enhancements to its Purview platform within the Microsoft Fabric ecosystem are designed to empower organizations with robust data security and governance, directly addressing the critical challenges of adopting artificial intelligence. These innovations focus on three core areas: identifying potential data risks, preventing the oversharing of sensitive information, and strengthening overall governance and data quality. By integrating Purview’s capabilities directly into Fabric, Microsoft provides a unified foundation for businesses to leverage their data for AI with greater confidence and control.
The integration between Microsoft Purview and Microsoft Fabric creates a cohesive environment for data security. This unified approach ensures sensitive information is protected, maintains clear visibility across the entire data landscape, and helps prepare data for responsible AI applications. As Darren Portillo, a Product Marketing Manager at Microsoft, notes, this integrated foundation allows companies to innovate confidently, knowing their data is both governed and trustworthy for AI initiatives.
A key focus is on discovering risks and preventing data from being shared too widely. Microsoft Purview brings a comprehensive suite of data protection tools to the Fabric environment. This includes Information Protection, Data Loss Prevention (DLP), Insider Risk Management (IRM), and Data Security Posture Management (DSPM). Specifically, DLP policies within Fabric can automatically identify sensitive data and trigger alerts. Administrators can use these policies to restrict access to sensitive data stored in various Fabric workloads, such as data warehouses, ensuring only authorized personnel can view it and significantly reducing the risk of exposure.
The platform also extends insider risk management to cover Fabric lakehouses. It uses built-in indicators based on user activity to spot potential risks like data exfiltration. Organizations can configure policies to detect suspicious actions, such as unusual data exports. A flexible pay-as-you-go model lets businesses track their usage and associated costs for these services.
Managing how data is exposed within AI interactions is another critical advancement. New preview features allow for risk detection within Copilot and AI agent interactions. Companies can now identify sensitive information contained in both user prompts and AI-generated responses. The system provides recommended actions to help mitigate these risks proactively. Furthermore, DSPM offers continuous data risk assessments, pinpointing Fabric assets that are sensitive or lack proper protection. These insights enable teams to apply necessary data labels or security policies to reduce potential exposure.
Beyond security, Microsoft Purview strengthens governance and improves data quality, which is essential for reliable AI outcomes. The Purview Unified Catalog offers a centralized, holistic view of all data assets, supporting strong enterprise governance. It helps teams discover and utilize data while adhering to governance standards required for AI projects. Built-in data quality tools allow organizations to measure, monitor, and enhance the reliability of their information, fostering trust in the data used for both decision-making and AI models.
For users of the Fabric OneLake catalog, Purview extends these governance capabilities, making them accessible to both data consumers and governance teams. The platform also introduces controlled workflows for data publishing. Within the Unified Catalog, data owners can manage the publication process for data products and business glossary terms, ensuring consistency and maintaining a well-curated data catalog.
To accelerate access to reliable data, new quality checks can be run on unmanaged data assets without the need to formally link them to data products. This capability enables faster, large-scale data validation, broadening access to high-quality data for various AI use cases across the organization.
(Source: HelpNet Security)
