What challenges did Deloitte face before AI implementation?

Protect PII, PHI, PCI, and other sensitive customer data from being stored in Dialogflow CX logs. Prevent sensitive data from propagating into downstream logging, analytics, and monitoring systems while still allowing it to be used in virtual agent responses.

What was the impact of the AI solution?

Provides a production-ready way to stop sensitive information leakage into logs and monitoring systems. Preserves the ability to use sensitive information in response generation while keeping non-sensitive log data intact.

Source

GCPExpandedEvidence: Low25/100

Deloitte & Google Cloud: Redacting PII in Dialogflow CX via Contact Center AI Security Settings and Cloud DLP

Use case typeContact center modernizationUpdated Jun 13, 2026

Deloitte Canada and Google Cloud describe a production-ready approach to prevent sensitive information from being stored in Dialogflow CX logs. The article explains that contact center conversations can include PII, PHI, PCI, and other confidential information in intent/form parameters, session parameters, webhook data, and fulfillment response messages, and that these values can propagate into Google Cloud Logging unless redacted at the source.

Organization: Deloitte
Industry: Professional Services
Location: Canada
Published: November 2022

Reported outcomes

Strategic outcomes

Risk & compliancePrevented sensitive data leakage into logsRisk & complianceRedacted sensitive conversation data at sourceNew product / capabilityEnabled safe use of sensitive data in responses

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: Deloitte
Provider: GCP
Maturity: Unknown
Linked source: Google Cloud Blog

No explicit deployment-stage evidence found.

Customer identity supportedSource describes one deploymentMaturity evidence evaluated

Primary read

Use case focus

Showing 3 of 3

1Data Privacy
2Security
3Contact Center Automation

Protect PII, PHI, PCI, and other sensitive customer data from being stored in Dialogflow CX logs.
Prevent sensitive data from propagating into downstream logging, analytics, and monitoring systems while still allowing it to be used in virtual agent responses.

Use built-in Dialogflow CX parameter redaction for intent and form parameters.
Use Cloud Data Loss Prevention inspection templates with Google Cloud Contact Center AI Security Settings to redact session parameters, webhook data, and fulfillment response messages before they reach Google Cloud Logging.
Use SSML mark tags to delimit sensitive spans so they can be redacted without affecting TTS behavior.
Implement the configuration through the Google Cloud Console, APIs, and Terraform.

Technologies

Google Cloud Contact Center AI Dialogflow CX Cloud Data Loss Prevention Google Cloud Logging Terraform Speech Synthesis Markup Language

Provides a production-ready way to stop sensitive information leakage into logs and monitoring systems.
Preserves the ability to use sensitive information in response generation while keeping non-sensitive log data intact.

Architecture

Dialogflow CX redaction is applied at multiple stages: built-in redaction for intent/form parameters, and CCAI Security Settings that reference a Cloud DLP inspection template to redact session parameters, webhook payloads, and fulfillment messages before they are published to Google Cloud Logging. SSML <mark> tags are used to preserve speech output while identifying sensitive spans. The configuration can be managed in Google Cloud Console or provisioned with Terraform using infrastructure-as-code.

Sources & evidence1

Evidence: Low25/100Evidence strength