The correct option is Perform code review and run security and license checks on Copilot generated code before promotion to production. This ensures the generated code meets quality, security, and compliance requirements before it reaches your Cloud Run service and connects to Cloud SQL.

This practice adds a human review step that can catch logic flaws, insecure patterns, secrets, and improper data handling that compilation and unit tests do not detect. Pairing review with automated scanning such as static analysis, dependency vulnerability checks, license compliance checks, and secret scanning establishes a reliable gate in continuous integration so only vetted changes are promoted to production within the planned timeline.

Adopt Copilot suggestions without changes if the code compiles and unit tests succeed is incorrect because compilation and unit tests do not validate security posture, licensing obligations, or the presence of hardcoded secrets. Relying on those signals alone invites vulnerabilities and compliance issues into production.

Turn off telemetry and feedback collection to address privacy concerns is incorrect because disabling telemetry does not improve the safety or compliance of the generated code. Responsible adoption focuses on review and scanning to manage risk rather than turning off data collection.

Depend on Cloud Armor and VPC Service Controls to offset any insecure code patterns is incorrect because these are perimeter and data exfiltration controls. They help with request filtering and service isolation but they do not remediate insecure application logic, injection flaws, or license violations introduced by the code itself.

Full AWS Practitioner Certification Question