SB 322

As substituted on 3/2/2026: - Creates the Educational Technology Regulatory Sandbox to enable time-limited pilot programs for evaluating educational technologies in public education, including artificial intelligence. - Requires applicants to meet required safeguards, including age-appropriate content, student data privacy compliance, educator oversight, parental notice and opt-out rights, educator training, outcome measurement, and completion of required safety testing. - Requires a vendor to conduct red-teaming exercises. - Prohibits vendors from deploying AI systems that provide sexual or harmful content, promote self-harm, manipulate students emotionally, simulate romantic or dependency-based relationships, discourage seeking help, or operate without educator supervision. - Prohibits a system from offering mental health diagnosis, counseling, or acting as a substitute for licensed professionals, and must direct students to crisis resources if they express suicidal or self-harm ideation. - Prohibits vendors from assigning final grades or making academic placement decisions without educator review, or providing full assignment answers that undermine learning. Provides that students retain ownership of AI-assisted work, and educators maintain final authority over grading, instructional decisions, academic records, and academic standards. - Requires disclosure to students when they are interacting with AI, that they can receive explanations of AI-generated outputs, request human review of decisions before they take effect, report concerns without retaliation, access education regardless of AI availability, and have their needs addressed by human educators. - Creates the Education Technology Advisory Council to advise state agencies on the implementation of AI pilot programs. - Requires that any statewide adoption of an AI system first be independently evaluated and receive specific legislative authorization. - The 3/2/2026 amendment renames the sandbox to apply to all educational technology. Creates an expedited application process for an AI system already approved for another sandbox participant. Grants rulemaking authority and creates an advisory committee.