Rampart: An On-Device ONNX Model for PII Detection and Client-Side Redaction
Summary
Rampart is a 14.7 MB ONNX token-classification model for on-device detection of personally identifiable information (PII) in text, released by National Design Studio. It serves as the client-side component of a defense-in-depth redaction system, running alongside a deterministic recognizer layer for structured identifiers. The model card documents only the released ONNX artifact, noting that alternative configurations (ELECTRA-small base, prefilter-off training variant, leaner data) were explored during model selection.
Source
Twitter / XRampart: An On-Device ONNX Model for PII Detection and Client-Side Redactionhuggingface.coKey quotes
· 5 pulledrampart is a 14.7 MB ONNX token-classification model that detects personally identifiable information (PII) in text before it leaves the user's device.
It is the on-device half of Rampart, a defense-in-depth client-side redaction system released by National Design Studio.
The shipped artifact runs alongside a deterministic recognizer layer that handles structured identifiers; together they form the complete system.
This card documents the released artifact only.
We're on a journey to advance and democratize artificial intelligence through open source and open science.
You might also wanna read
Local-First Reversible PII Scrubber for AI Workflows Using ONNX and Regex
The article discusses a technical solution to the privacy-translation paradox in AI workflows, where teams need to translate user content us
medium.com·6mo ago
Microsoft open-sources RAMPART and Clarity: Safety tools for AI agent development
Microsoft is open-sourcing two new tools — RAMPART and Clarity — designed to help engineers integrate safety measures into the AI agent deve
SIR-Bench: A Benchmark for Evaluating Autonomous Security Incident Response Agents
Researchers introduce SIR-Bench, a comprehensive benchmark for evaluating autonomous security incident response agents. The benchmark consis
arxiv.org·2mo ago
Study Reveals Invisible Manipulation Vulnerability in AI Financial Advisory Systems That Evades All Current Detection Methods
This paper identifies and empirically validates an invisible manipulation channel in AI-assisted financial advisory systems, specifically at
arxiv.org·13d ago
AI Assistant Companies Shift to Advertising Models While Developing Always-On Surveillance Hardware
The article examines how major AI assistant companies like OpenAI are transitioning to advertising-based business models while simultaneousl
juno-labs.com·4mo ago
MiniCPM 4.0: Ultra-Efficient Open-Source AI Models for On-Device Deployment
MiniCPM 4.0 is an ultra-efficient, open-source AI model family designed for on-device deployment, featuring significant speed improvements o
Product Hunt·9mo ago
Comments
Sign in to join the conversation.
No comments yet. Be the first.