Toxicity

Harmful, offensive, or abusive content in model output.

Toxicity detection is part of responsible deployment.

Related terms

Back to Guidelines for Responsible AI