Dinghao Wu

On the Safety of Open-Sourced Large Language Models: Does Alignment Really Prevent Them From Being Misused?