DEFINITION
Privacy Layer for AI
A control point between a user and an AI service that detects, transforms, or blocks sensitive data before it reaches the model.
In depth
A privacy layer for AI sits at the boundary between a user's workflow and a third-party model. Its job is narrow: inspect outbound prompts and uploads, recognise sensitive content, and decide whether to allow, mask, tokenize, or block. The layer is independent of which AI provider is in use, so the same controls cover ChatGPT, Claude, Gemini, Grok, and any future model. It is to AI what a data loss prevention (DLP) gateway has been to email and file transfers — but redesigned to work on free-form text and to run locally rather than as a corporate proxy.
Examples
- Locke flags privileged client material in the ChatGPT composer.
- A desktop agent that masks API keys before they reach a coding assistant.
- An enterprise gateway that routes prompts through tokenization rules per business unit.
How Locke handles privacy layer for ai
Locke is a privacy layer for AI. Detection and masking are coming to Locke, the Sonomos desktop app, for native AI clients. Both layers run 100% on-device.
See the productRelated terms
- Prompt Leakage
The unintended disclosure of sensitive information through the content of a prompt sent to an AI model.
- Zero-Knowledge Processing
An architecture where the service provider has no technical ability to read user data, because all sensitive processing happens on the client.
- Tokenization
Replacing a sensitive value with a non-sensitive token that maps back to the original through a separate, controlled lookup.
