Taplio - Build a growth routine on LinkedIn
Mini toolsPrivacy Filter
By OpenAI
Privacy Filter is an open Apache 2.0 privacy-span detection model from OpenAI for redaction and data minimization. The Hugging Face card says it is a smaller gpt-oss-like checkpoint converted into a bidirectional token classifier with constrained Viterbi decoding, using 1.5B total parameters with 50M active parameters. It detects 8 privacy categories, including names, emails, phones, addresses, account numbers, dates, URLs, and secrets, and is intended for fast, tunable, on-premises PII masking rather than blanket anonymization guarantees.
New
Text
Gen 7
Overview
Privacy Filter is OpenAI’s bidirectional token-classification model for detecting and masking personally identifiable information in text. It is built for high-throughput data sanitization workflows, can run on-premises, supports a 128,000-token context window, and is designed to label text in a single forward pass rather than token-by-token generation.
About OpenAI
OpenAI is a technology company that specializes in artificial intelligence research and innovation.
Industry:
Artificial Intelligence
Company Size:
4500
Location:
San Francisco, California, US
Website:
openai.com
Tools using Privacy Filter
No tools found for this model yet.
Last updated: April 22, 2026
