First-pass redaction before sharing
Mask common contact and identifier formats, then inspect the text for anything remaining.
Mask structured patterns such as email and phone numbers, with optional BERT NER detection for candidate names, organizations, and locations.
Mask common contact and identifier formats, then inspect the text for anything remaining.
Enable AI detection to add possible names and organizations, then review the masked output.
Text is processed in the browser with regular expressions and, when selected, the NER model. With your consent, model files may be downloaded from Hugging Face and cached in browser storage, while your input is not sent to the model provider.
Pattern masking works without a model. Name and organization detection needs about 181 MB of model files and cannot guarantee detection of every sensitive item, so manual review is essential.