Glossary · AI in customer service
What is PII masking for AI?
In short
PII masking replaces personal and sensitive details, such as card numbers, bank accounts, ID numbers, email addresses and phone numbers, with placeholders before a message is sent to an AI model. The model can still understand and answer the question without ever seeing the real values.
Reviewed 9 November 2026. How it works in Axonave: Axonave AI.
Placeholders in, real values kept
PII means personally identifiable information. Customers type it into chats all the time: “my card ending 4417 was charged twice”, an IC number to confirm who they are, a phone number to call back.
Masking finds those details by their pattern and swaps each one for a label, so the model reads “my card [CARD_1] was charged twice”. The sentence keeps its meaning, so the answer is as good. The real values stay in your own system, where your team can still see them.
Masking is one control among several. It covers details with a recognisable shape; it doesn’t make it safe to paste a whole customer file into a prompt. Send the AI only what the task needs.
- Found by pattern: cards, bank accounts, ID numbers, secrets
- Replaced with labels the model can still reason about
- The original kept for your team, never sent to the model
PII masking for AI: common questions
Will the AI still understand the question?
Yes. The placeholder keeps the sentence’s shape (“my card [CARD_1] was charged twice”), which is all the AI needs to find the right article and answer.
Does masking hide the card number from our agents?
No. Masking applies to what’s sent to the AI. The stored conversation and your team keep what the customer wrote.
Can we switch masking off for card numbers?
No. Cards, bank accounts, IC and passport numbers and secrets are always masked. Emails and phone numbers are masked by default.
See it working with your own customers.
We reply within one business day by email or WhatsApp.
