LlamaGuard-2

Sentra-Guard system achieves 99.96% detection rate against adversarial LLM prompts

Researchers have developed Sentra-Guard, a real-time system designed to defend against adversarial prompts targeting large language models. The system employs a hybrid approach combining semantic embeddings with transfo…