lighthouz-ai-a3db3c0d·1 events·first seen Aliases: Lighthouz AI
Hugging Face and Lighthouz AI have launched the Chatbot Guardrails Arena, a new evaluation platform focused on assessing safety guardrails in conversational AI systems. The arena uses human preference-based evaluation to benchmark how well different chatbot guardrail implementations resist unsafe or policy-violating outputs. This fills a gap in existing evaluation infrastructure, which has largely focused on capability rather than safety constraint enforcement.