Comment by UNICEF

For certain high-risk categories, such as sexually explicit or ‘romantic’ AI chatbots/companions, companies should be required to prevent access by children.

AI chatbots and companions should be subject to mandatory pre-deployment testing and reliable post-deployment monitoring by independent, qualified third parties, with results reported to the regulator/oversight body.

Researchers, in coordination with statutory child rights institutions, regulators, and civil society, should establish public, open repositories of safety incidents, research findings and regulatory evaluations to inform policy, enforcement and techn...