AgentAlign: Navigating Safety Alignment in the Shift from Informative to Agentic Large Language Models

Open in new window