OpenAI Open Source Privacy Filter Model

OpenAI has released its OpenAI Privacy Filter model, designed for detecting and editing personally identifiable information (PII) in text. The model has 1.5 billion parameters, supports 128K contexts, and can run locally. Based on a bidirectional token classification architecture, it can identify eight types of privacy information, including name, address, email, phone number, account, and password, achieving a 96% F1 score on the PII-Masking-300k benchmark.