OpenAI releases 1.5B Privacy Filter model
openai released privacy filter, a 1.5b-parameter model for pii redaction under apache 2.0. it features a 128k context window and achieves sota on the pii-masking-300k benchmark. three new gradio-based apps demonstrate the model for documents, images, and clipboard text.
OpenAI released Privacy Filter, a 1.5B-parameter model licensed under Apache 2.0 for detecting and redacting Personally Identifiable Information (PII). The model utilizes 50 million active parameters, supports a 128,000-token context window, and achieves state-of-the-art performance on the PII-Masking-300k benchmark. It identifies eight categories of sensitive data including private persons, emails, phone numbers, and account numbers.
The release includes three reference applications built on gradio.Server to demonstrate the model's capabilities: a Document Privacy Explorer for highlighting PII in PDFs and DOCX files, an Image Anonymizer for redacting screenshots with editable black bars, and SmartRedact Paste for sharing redacted text via public URLs.