Deepseek’s AI model proves easy to jailbreak – and worse Posté par admin | Jan 31, 2025 | Techno | 0 | In one security firm’s test, the chatbot alluded to using OpenAI’s training data. Aller à la source