Why it matters
New York’s signed S9051B takes effect January 1, 2027. It restricts defined unsafe features for minors in covered conversational AI systems, with specific exceptions for limited-purpose and internal uses. The law also sets age-assurance safeguards and limits use and retention of age-check data. Coverage depends on the statutory definitions and actual product behavior.
My takeaway: Map scope and exceptions before translating restricted behaviors into tests. Check applicable age-assurance rules; the interim provisions require multiple methods, including one without government ID or one preserving anonymity from the operator. Test circumvention and immediate deletion of age-check data, documenting any legally required retention. Recheck implementing rules before release.