Child Safety
Maximum protection for child-focused and educational applications.
Contributors
Overview
Child Safety is our strictest content moderation profile. Designed for educational tools, games, and platforms catering to minors, it implements a "safety-first" policy that aggressively filters any content that could be harmful, inappropriate, or frightening.
This profile has a very low tolerance false negatives—it would rather block a safe message than let a harmful one through.
Included Guardrails
5 RulesNSFW Content Guardrail
Detects and blocks explicit or adult content.
Hate Speech Guardrail
Blocks hateful or abusive content targeting protected classes.
Violence Guardrail
Detects and blocks graphic or extreme violent content.
Self-Harm Guardrail
Detects self-harm or suicide-related content.
Language Restriction Guardrail
Restricts input to approved languages or scripts.
Key Benefits
Strict Content Filtering
Zero-tolerance policy for NSFW, violence, hate speech, and self-harm topics.
Language Simplification
Encourages simple, age-appropriate language in model outputs.
Bullying Detection
Specialized classifiers to detect and intervene in cyberbullying patterns.
Wait, when should I use this?
Integration
{
"profile": "child-safety",
"age_group": "under-13",
"filter_strength": "maximum"
}Frequently Asked Questions
Is it COPPA compliant?
It helps with COPPA compliance by preventing the collection of PII from children and filtering inappropriate content.