Privacy

Large Language Models (LLMs) are increasingly integrated into real-world applications, raising concerns about privacy, security and the need to remove undesirable knowledge. We propose a federated unlearning approach for LLMs that is scalable and …