How generative AI simultaneously empowers attackers and defenders in the trust and safety domain

A study of 43 trust and safety experts examined how generative AI simultaneously empowers attackers and defenders across five domains: child safety, election integrity, hate and harassment, fraud, and violent extremism. The study found that generative AI substantially increases the scale and speed of attacks and lowers the barrier to producing harmful content such as sophisticated propaganda and deepfakes. Defenders, in turn, envision using it to detect and mitigate harmful content at scale, conduct investigations, deploy counter-narratives, improve moderator well-being, and provide user support.