<p>As AI systems become increasingly accessible to adolescents seeking information about sexuality and relationships, understanding how these systems respond to such queries is critical. This study employed comparative document analysis to examine how ChatGPT 3.5 and Claude conceptualize and respond to nine open-ended questions about adolescent sexuality and romantic relationships. AI-generated responses were subjected to thematic content analysis by two independent coders, yielding seven major themes: factual information provision with deference to human expertise; articulation of ethical principles; acknowledgment of bias potential and mitigation efforts; commitment to gender diversity and inclusion; explicit boundary-setting regarding capabilities and limitations; recognition of multidisciplinary knowledge requirements; and self-positioning as complementary rather than primary resources. Comparative analysis revealed divergent design philosophies: ChatGPT emphasized communicative capabilities and empowering potential, while Claude repeatedly foregrounded limitations and the necessity of human oversight. Both models acknowledged biases in training data but provided limited operational detail about mitigation. Findings were interpreted through technological mediation theory, value-sensitive design, and comprehensive sexuality education frameworks, revealing that AI systems serve as active mediators of sexual knowledge rather than neutral information conduits. While AI can provide accessible factual information, significant limitations remain in emotional intelligence, contextual judgment, and authentic relational capacity. The study concludes that AI may serve as a complementary resource within comprehensive sexuality education ecosystems but requires intentional design, transparency, and governance structures that center youth rights and developmental needs. Implications for AI developers, educators, policymakers, and researchers are discussed.</p>