> ## Documentation Index
> Fetch the complete documentation index at: https://help.aichat.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Content Redaction
Content redaction automatically detects and hides sensitive customer information such as email addresses, phone numbers, and credit card details before messages are stored or processed.
Content Redaction is currently available for Enterprise tier only.
**Before you start, ensure that:**
* You have administrator access to configure Content Redaction
* You understand whether original sensitive data should be viewable or permanently removed
# How Content Redaction Works
When a customer sends a message containing sensitive data, AiChat performs the following in real time:
1. **Detection:** Identifies supported data patterns.
2. **Redaction:** Replaces the value with a placeholder (e.g., `[EMAIL REDACTED]`).
3. **Storage:** Saves only the masked version in the conversation history
4. **AI Processing:** Only the masked content is sent to AI providers to ensure data privacy.
## Supported Data Types
AiChat includes built-in detection for the following:
| **Data Type** | **What it Detects** | **Validation Method** |
| :------------ | :------------------------------- | :---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Email Address | Standard email formats | Email must contain a local-part before the @, and a valid domain name. |
| Phone Number | International phone formats | Phone number must include country code in front to classify as international phone format |
| Credit Card | Visa, Mastercard, AMEX, Discover | Numbers that start with a known issuer prefix AND pass Luhn validation: - Visa`4xxx`13–16 digits
- Mastercard`51xx`–`55xx`16 digits
- Amex`34xx` or `37xx`15 digits
- Discover`6011` or `65xx`16 digits
|
**Custom Patterns:** Businesses can create custom redaction rules using regular expressions (regex) to mask organization-specific data like Employee IDs or account numbers.
## Data Handling Options
Administrators can choose between two policies for handling original sensitive data:
**Allow Viewing Original Data**
* Behavior: Original values are encrypted and stored.
* Agent Access: Support Agents can click a masked token to reveal or copy the original value.
* Use Case: Recommended for support teams that must verify customer details.
**Permanently Remove Original Data**
* Behavior: Original values are discarded immediately and cannot be recovered.
* Access: No reveal option is available to anyone.
* Use Case: Best for organizations with strict compliance or data minimization requirements.
* Warning: Once data is removed under this policy, it cannot be restored, even if settings are changed later.
# Configure PII Masking
### Enable Masking
1. Navigate to **Settings > Content Redaction.**
2. Toggle **Enable Content Redaction**.
3. Select specific data types to detect (e.g., enable Email but disable Phone).
4. Select your Data Handling Policy.
5. Click **Save.**
### **Add Custom Masking Patterns**
1. In Content Redaction, click **Add Pattern.**
2. Provide a Pattern Name, the Regex Pattern, and a Test Input.
3. Review the preview and **Save.**
# Key Considerations
* PII Masking only applies to customer messages and free-text responses collected through Collect nodes. Bot and agent responses are not redacted.
* Data submitted through Collect with Form is not redacted. Use Collect with Form when you need to collect sensitive information accurately.
* Customers using AiChat Web Chat will see sensitive information masked after the message is sent.
* Customers using social messaging platforms such as WhatsApp will continue to see their original unmasked messages on their own devices, as the conversation history is stored outside of AiChat.