AI safety mechanisms are falling short with newer multimodal AI models like GPT-4o June 24, 2024 by admin Tweet A study has found that multimodel AI models perform poorly at giving safe responses when users give multimodal inputs such as an image and text together. The new SIUO benchmark was made as a result. Read more… Neowin Related Posts:Microsoft Azure AI Content Safety adds a new way to…MediaTek's flagship Dimensity 9400 to support…OpenAI's GPT-4o mini surpasses Gemini 1.5 Flash and…Inside GPT-4o: OpenAI's System Card details safety…Microsoft releases Phi-3.5 family of models that…