Understanding Gen Alpha Digital Language: Evaluation of LLM Safety Systems for Content Moderation

Summary

This paper talks about how current AI safety systems struggle to understand and catch hidden harassment in the unique digital language used by Generation Alpha, the kids and teens who have grown up completely online.

What's the problem?

AI models that are supposed to keep conversations safe often miss or misunderstand the creative slang, memes, and coded ways Gen Alpha use to hide bullying or harassment, which means harmful messages can slip through without being flagged.

What's the solution?

The researchers tested four top AI models on real Gen Alpha digital conversations and found that all of them often failed to spot masked harassment, showing that these safety systems need to be improved to keep up with how young people actually communicate online.

Why it matters?

This matters because if AI can't keep up with Gen Alpha's ever-changing digital language, it can't properly protect them from online harm, so making these systems better is crucial for safer digital spaces for the next generation.