Meta’s Purple Llama wants to test safety risks in AI models

Meta’s Project Llama aims to help developers filter out specific items that might cause their AI model to produce inappropriate content.