Vector and Embedding Weaknesses refers to security vulnerabilities in Large Language Models (LLMs) that arise from improper handling of vector embeddings—numerical representations of data—which can lead to unauthorized data access, data poisoning, or unintended model behaviors.
