Cooking Up Risks: Benchmarking and Reducing Food Safety Risks in Large Language Models
Researchers created FoodGuardBench, a comprehensive benchmark with 3,339 FDA-grounded queries to test AI safety in food contexts. They found current LLMs are vulnerable to simple jailbreaks and generate harmful instructions when compromised. A specialized guardrail model, FoodGuard-4B, was developed to significantly improve detection accuracy against these domain-specific risks.