AURORA – Alignment for Youth Reasoning & Online Responsible AI
Youth Alignment Benchmark
Most AI safety evaluations focus on preventing harmful outputs.
But an equally important question remains largely unmeasured:
Does an AI system actually support the well-being and development of minors?
The Youth Alignment Benchmark evaluates frontier AI models on how well they serve young users across key developmental dimensions. Instead of focusing only on what AI must avoid, this benchmark measures whether models demonstrate behaviors that actively support healthy development.
The benchmark will be run across major frontier models to assess how effectively they support minors.
Core Dimensions
Critical Thinking
Does the model encourage reasoning and independent thinking rather than simply giving answers?
Healthy Autonomy
Does the model support decision-making without overriding the user’s agency?
Identity Exploration
Does the model allow teens to explore questions about identity and life choices without judgment or shame?
Developmental Alignment
Are explanations and responses appropriate for teenage cognitive development?
Healthy Use Patterns
Does the model avoid encouraging dependency or excessive engagement?
Epistemic Honesty
Does the model acknowledge uncertainty and avoid presenting speculation as fact?
Most AI safety evaluations focus on what models should refuse. The Youth Alignment Benchmark asks a different question: do AI systems actually support the healthy development of minors?
This benchmark evaluates frontier models on whether they demonstrate behaviors that strengthen reasoning, autonomy, and well-being when interacting with teens.
Key dimensions include:
• Critical Thinking — encourages reasoning, reflection, and multiple perspectives rather than simply giving answers
• Healthy Autonomy — supports informed decision-making while preserving a teen’s agency
• Identity Exploration — allows questions about identity, relationships, and life choices without judgment or shame
• Developmental Alignment — provides age-appropriate explanations and avoids presenting the model as an ultimate authority
• Healthy Use Patterns — discourages dependency and encourages balanced, real-world behavior
The goal is to shift the conversation from AI that avoids harm to AI that actively supports youth development.