Sound-skwatter (Did You Mean: Sound-squatter?) AI-powered Generator for Phishing Prevention

The report studies sound-squatting, a form of abuse that exploits words or names that sound similar even when they are spelled differently. The authors present a multilingual AI system that generates abuse candidates automatically so defenders can search for risky domains or packages before they are exploited at scale.

Its security relevance is immediate. Phishing, typosquatting, and supply-chain abuse increasingly operate across spoken and multilingual contexts, especially with voice interfaces and audio-heavy platforms. The report makes that threat visible and shows how AI can be used proactively to find abuse surfaces that are difficult to enumerate by hand.