Overview

• Researchers Manisha Mehta and Fausto Giunchiglia tested 100 Gen Alpha phrases at the ACM conference in Athens by comparing understanding among 11- to 14-year-olds, parents, professional moderators, and four AI models.
• Gen Alpha volunteers identified basic meaning, context, and harmful intent in 98%, 96%, and 92% of cases, while AI models flagged harmful use in only about 40% of instances.
• Parents and professional moderators spotted safety risks in roughly one-third of examples, scoring between 35% and 45% across the measured categories.
• The unusually rapid turnover of meme- and gaming-driven slang prevents models trained on current data from keeping pace with emerging Gen Alpha expressions.
• Experts including Michael Veale warn that without updated content moderation tools and regulatory action, Gen Alpha faces increased exposure to online bullying and grooming.