DeepSeek 'highly vulnerable' to generating harmful content, study reveals
DeepSeek took the world by storm this week. However, new research reveals the Chinese AI model has serious ethical and security flaws, including being 11 times more likely to generate harmful output than OpenAI’s o1.