基于政策的安全性评估：20个大语言模型
Policy-Grounded Safety Evaluation of 20 Large Language Models
AI总结：
本文提出Aymara AI平台，用于生成和管理定制化的政策导向安全性评估。通过评估20个商业大语言模型在10个现实安全领域中的表现，揭示了模型在不同领域中的显著性能差异，强调了构建可扩展、可定制工具的重要性。
摘要：
随着大语言模型（LLMs）越来越多地融入现实应用，可扩展且严格的安全性评估至关重要。本文介绍Aymara AI，一个程序化平台，用于生成和管理定制化的、基于政策的安全性评估。Aymara AI将自然语言安全政策转换为对抗性提示，并使用经过人类判断验证的AI评分者来评分模型响应。我们通过Aymara LLM风险与责任矩阵展示了其能力，该矩阵在10个现实安全领域评估了20个商业可用的LLM。结果揭示了广泛的性能差异，平均安全分数从86.2%到52.4%不等。虽然模型在已建立的安全领域如虚假信息（平均=95.7%）表现良好，但在更复杂或未明确规定的领域，如隐私与身份冒充（平均=24.3%），表现持续失败。方差分析确认了模型和领域之间的安全分数存在显著差异（p < .05）。这些发现突显了LLM安全性的不一致性和情境依赖性，并强调了构建可扩展、可定制的工具如Aymara AI来支持负责任的AI开发和监管的重要性。
英文摘要：
As large language models (LLMs) become increasingly integrated into real-world applications, scalable and rigorous safety evaluation is essential. This paper introduces Aymara AI, a programmatic platform for generating and administering customized, policy-grounded safety evaluations. Aymara AI transforms natural-language safety policies into adversarial prompts and scores model responses using an AI-based rater validated against human judgments. We demonstrate its capabilities through the Aymara LLM Risk and Responsibility Matrix, which evaluates 20 commercially available LLMs across 10 real-world safety domains. Results reveal wide performance disparities, with mean safety scores ranging from 86.2% to 52.4%. While models performed well in well-established safety domains such as Misinformation (mean = 95.7%), they consistently failed in more complex or underspecified domains, notably Privacy & Impersonation (mean = 24.3%). Analyses of Variance confirmed that safety scores differed significantly across both models and domains (p < .05). These findings underscore the inconsistent and context-dependent nature of LLM safety and highlight the need for scalable, customizable tools like Aymara AI to support responsible AI development and oversight.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
