Fetching the paper…

LLM Safeguard is a Double-Edged Sword: Exploiting False Positives for Denial-of-Service Attacks · Around