AI-assisted Online Child Protection

摘要

No one can deny the internet’s great effect on the progress that we see around us.  Despite all the benefits of the internet, it has its dark side, when misused the internet  represents a threat, especially for children. Lacking child internet safety is a  hazardous problem that people usually ignore or simply underestimate. According  to the Center for Cyber Safety and Education, 53% of kids in grades 4-8 revealed  their number to a stranger, and about 11% have met a stranger[1]. These statistics  are skyrocketing, which is scary. In this research paper, a prototype for an online  child safety software that utilizes AI to identify obscene photos. The reasoning  behind writing this research is to explore one of the ways that a child protection  software can be created and to explore and empathize the importance of  maintaining child privacy while protecting them. The software here will mainly  focus on limiting the threats from these sources. The algorithm is designed to block  all unwanted websites with any unfriendly or aggressive content that does not suit  children. Unblocked websites are continuously analyzed, where users receive the  results of these updates periodically. As for the videos, AI can recognize and  analyze videos to ensure the safest online experience for the child. The chatrooms  are the last approach we are taking into consideration in our project. We are  scanning the chats to recognize any obscene or immoral content, when any  prohibited content is detected the parents are alerted at once.