Our paper on risk-averse model uncertainty for safe RL appeared at NeurIPS 2023.