Safe Space Alliance：2025年对话式AI智能体安全评级（CAASR）报告（英文版）

In developing the inaugural Conversational AI Agent Safety Rating (CAASR) Report, research began with a focus on supporting LGBTQI+ communities. It then broadened to encompass other marginalised groups, and has since evolved to address the needs of all people, reflecting our commitment to universal understanding and inclusion. Conversational AI agents stand at a pivotal juncture in their development, offering remarkable potential to serve as tools for emotional support, intellectual explorati