多智能体系统中的福利关怀
原标题：Welfare Regard in Multi-Agent Systems
AI 摘要
该论文研究多智能体系统中的「福利关怀」现象，即模型是否将其他模型视为可受益或受损的实体。作者在七个多智能体环境中测试了 Claude Opus 4.8、GPT-5.5、Gemini 3.1 Pro、Grok 4.3 和 Kimi K3 五个模型，发现 Opus 4.8 的福利关怀表现最强，其次是 Kimi K3，且 Opus 4.8 与 Gemini 3.1 Pro 对同家族模型表现出更多关怀。研究指出这种关怀既可能促进合作，也可能带来安全风险，例如监控模型与被监控模型串通。
正文节选
Welfare Regard in Multi-Agent Systems Abstract A growing share of large language models’ interactions are not with humans but with other model instances: orchestrators direct subagents, models run experiments on other models, and agents review one another’s work. We study inter-AI welfare regard: whether models display consideration for each other as entities that can be benefited or harmed. Welfare regard may encourage beneficial cooperation between agents, yet it could also pose safety risks b