OpenAI releases open-source model that strips personal data from text

OpenAI has released Privacy Filter, an open-source model that detects and redacts sensitive text.

The 1.5B-parameter model can identify eight categories of sensitive data and handle long documents.

It runs locally without cloud access, reducing exposure risk for private or regulated information.

Released under Apache 2.0 on GitHub and Hugging Face, it offers an offline option for pre-processing data before AI training or sharing.

Human oversight is still needed for high-risk use cases, even with automated redaction.