Glossary · Safety

Toxicity Classifier

A model trained to detect harmful, abusive, or policy-violating content, commonly used as a guardrail layer to filter inputs or outputs in production AI applications.