ToolHazard: Scaling Adversarial Environments for Security Evaluation and Alignment of LLM-based Agents

Part of

Self-Driving Car Security Evaluation