GLiNER2-Guardrails-PII-Multi

A unified multilingual model for PII detection & redaction and LLM safety moderation. Supports EN, FR, ES, DE, IT, PT, NL. One 0.3B checkpoint, two capabilities.

Model · GitHub · PII Paper · GLiGuard Paper

Detect 42 types of personally identifiable information in text. Entities are highlighted and can be redacted with [LABEL] placeholders.

Input Text

PII Labels (comma-separated)

Threshold

Lower = more detections (may increase false positives).

0 1

Show redacted text

Detected Entities

Redacted Text

Raw Output (JSON)

Examples