OpenAI sketches six pillars for teen AI safety

OpenAI published a six-pillar Australian Youth Safety Blueprint, pairing privacy-preserving age assurance with under-18 defaults, crisis-response protocols, and parental controls for users identified as 13–17. The primary blueprint asks policymakers to require independent assessments and plain-language summaries of how providers implement teen safeguards, but offers no outcome data that would make those controls a safety certification. Builders shipping youth-facing AI should treat this as a policy template and evidence request, not a pass; China’s 345 million-user companion shutdown shows how quickly regulators can turn emotional-AI design choices into product constraints.