RELATING TO ARTIFICIAL INTELLIGENCE FOR THE PROTECTION OF MINORS.
What changed between versions
Added a new definition for 'material risk of harm to minors' covering self-harm and unauthorized data disclosure, while removing definitions for 'emotional dependency' and 'relational attachment.'
Changed prohibited AI behaviors to specifically ban representations that the AI has independent emotions, consciousness, or attachment, and to ban claims of providing professional medical or therapeutic services.
Mandated that providers create and publish a 'minor protection plan' on their website, including risk assessments and mitigation strategies.
Required providers to submit summaries of their risk assessments and mitigation steps to the office of consumer protection before deploying new AI systems.
Removed the provision allowing private citizens to sue for violations of the bill.
Added a new subsection clarifying that the bill's duties are cumulative with other laws, though private actions are still barred.
Increased civil penalties for violations to up to $15,000 per violation per day.