OpenAI releases Privacy Filter on Hugging Face — 1.5B-parameter open PII redaction model

OpenAI quietly released Privacy Filter on Hugging Face under Apache 2.0: a 1.5B-parameter bidirectional token-classification model with 50M active parameters, purpose-built for detecting and redacting personally identifiable information at the token level.