This paper proposes strategies to increase trust in AI for railway applications by focusing on robustness, Operational Design Domain (ODD), and explainability. It suggests that existing algorithms and methods from non-railway actors can be applied to meet railway standards and regulations.