SnowPro Advanced: Security Engineer Question 405
Single answer5.2 Leverage Snowflake Cortex AI to enhance data security.A security engineering team at a healthcare company wants to reduce the manual effort required to identify sensitive information in newly ingested claims notes stored in Snowflake. They want to use Snowflake-native AI capabilities to help classify free-text content so they can drive downstream masking and governance decisions, while keeping the data inside Snowflake. Which approach is the MOST appropriate?
- A
Use Snowflake Cortex AI functions to analyze the unstructured text in the claims notes and generate classifications or labels that can be stored in Snowflake tables, then use those results to support security controls such as tagging, masking, or governance workflows.
- B
Use network policies to inspect the contents of claims notes at query time and automatically detect PHI in the result set before the data is returned to users.
- C
Use Tri-Secret Secure so Snowflake Cortex can natively classify sensitive text without requiring any SQL processing or governance objects.
- D
Use row access policies alone to identify PHI entities embedded inside free-text notes and dynamically rewrite the note contents to remove sensitive terms.
Show answer and explanation
Correct answer: A
Explanation
The best answer is to use Snowflake Cortex AI to analyze text directly within Snowflake and store the resulting classifications or indicators for downstream security and governance enforcement. In practice, organizations often combine AI-driven analysis of free text with Snowflake governance features such as tags, masking policies, and policy-based access controls. The key distinction is that Cortex provides AI/LLM capabilities for understanding content, while features such as network policies, row access policies, and encryption controls address different parts of the security model. Snowflake documentation and best practices distinguish clearly between content analysis capabilities and governance enforcement mechanisms: AI can help identify sensitive content, but enforcement still relies on established security features such as tagging, masking, and access policies.
- A. Correct.
Correct. Snowflake Cortex AI can be used to analyze text stored in Snowflake and produce summaries, labels, classifications, or extracted signals that help identify sensitive content. In a security workflow, those outputs can be persisted and then used to drive governance actions such as applying tags, informing masking strategies, or supporting review processes. This matches the requirement to keep data inside Snowflake and reduce manual review of free-text fields.
- B. Incorrect.
Incorrect. Network policies control which client IP addresses can connect to Snowflake; they do not inspect query payloads or classify data content. This option confuses perimeter access control with data content analysis.
- C. Incorrect.
Incorrect. Tri-Secret Secure is a key management and encryption-related capability, not an AI-based data classification tool. It does not inspect text or classify PHI. This distractor reflects a common misconception that stronger encryption features also provide data discovery or classification.
- D. Incorrect.
Incorrect. Row access policies control which rows a role can see based on conditions, but they do not perform NLP-style identification of PHI inside free-text columns and do not rewrite text contents for sanitization. They are useful after data has been classified, not for performing the classification itself.