OpenAI’s new Privacy Filter runs on your laptop so PII never hits the cloud

OpenAI launched Privacy Filter, a bidirectional token-classification model for detecting and masking common PII in long text.

The model runs locally with a 128,000-token context window, letting it scan lengthy documents, logs, and filings in one pass.

OpenAI says it aims for strong benchmark performance while giving developers a faster, context-aware way to redact sensitive data on-device.

The approach can reduce privacy risk by keeping more PII processing on laptops or in browsers instead of sending text to the cloud.