Smart redaction
Smart Redaction is a privacy feature in Elephas that automatically hides your personal and sensitive information before any of your data is sent to cloud AI models like ChatGPT, Claude, or Gemini.
What is Smart Redaction?
Smart Redaction is a privacy feature in Elephas that automatically hides your personal and sensitive information before any of your data is sent to cloud AI models like ChatGPT, Claude, or Gemini.
It works silently in the background — replacing things like names, emails, and financial details with anonymous labels, then putting the real information back into the AI's response on your Mac. The cloud AI never sees your real data.
Why does it matter?
When you use AI, anything you type or share — bank details, medical records, legal documents, client information, card numbers — normally travels to an online server. That data could be logged, stored, or used to train future models.
Smart Redaction prevents this by keeping your sensitive information on your device, while still letting you enjoy the full power of the best cloud AI models.
How does it work?
- You type a message or ask Elephas to work with a document.
- Elephas scans the text on your Mac and detects private information.
- Each piece of private data is replaced with an anonymous label (e.g.
PERSON_1,EMAIL_1,DATE_1).
- The anonymized text is sent to the cloud AI.
- The AI responds using the same labels.
- Elephas replaces the labels with your real information before showing you the answer.
Example:
Original: "Send a follow-up to Sarah Johnson at sarah.j@lawfirm.com about the March 15th meeting."
What information is protected?
Smart Redaction covers a wide range of sensitive data:
Personal Identity
- People's names
- Company and organization names
- Physical addresses and locations
Contact Details
- Email addresses
- Phone numbers (US, UK, European, and Indian formats)
Financial Information
- Credit and debit card numbers
- Bank account numbers
- IBANs and SWIFT codes
- Bank routing numbers
- Currency amounts
Government IDs
- Social Security Numbers (US)
- Tax IDs
- Passport numbers
- Driver's license numbers
- UK National Insurance numbers
- Indian PAN, Aadhaar, and GSTIN
Medical & Health Information
- Medical record numbers (MRN)
- Health insurance policy IDs
- NPI numbers
- NHS numbers
Other
- Dates
- URLs
- IP addresses
How do I know it's working?
After each AI message, you'll see a small Smart Redaction badge showing exactly what was protected — for example, "3 names, 1 email redacted."
You can also visit the Dashboard to view your overall protection activity over time.
How do I configure Smart Redaction?
Smart Redaction is on by default — no setup needed. But you have full control over how it behaves for each Brain.
Global Settings
Go to Preferences → Privacy to turn Smart Redaction on or off for the entire app.
Per-Brain Settings ( Only For Pro plus Users )
Open any Brain's settings to choose one of three modes:
Mode | What it does | Best for |
Use Global | Follows your master toggle in Preferences | Most everyday Brains |
Always On | Forces protection even if the global toggle is off | Highly sensitive Brains — client work, legal, medical, financial |
Off for this Brain | Disables redaction for this Brain only | Public or non-sensitive data where you want maximum AI accuracy |
Your settings sync across all your devices via iCloud.