SAGE: A Generic Framework for LLM Safety Evaluation
This work addresses the need for more robust and context-specific safety testing in AI deployment, though it is incremental as it builds on existing evaluation methods by adding modularity and dynamic interactions.
The authors tackled the problem of inadequate safety evaluation for LLMs in real-world applications by introducing SAGE, a modular framework for dynamic harm evaluation, which revealed that harm increases with conversation length and model behavior varies significantly with user personalities and policies.
As Large Language Models are rapidly deployed across diverse applications from healthcare to financial advice, safety evaluation struggles to keep pace. Current benchmarks focus on single-turn interactions with generic policies, failing to capture the conversational dynamics of real-world usage and the application-specific harms that emerge in context. Such potential oversights can lead to harms that go unnoticed in standard safety benchmarks and other current evaluation methodologies. To address these needs for robust AI safety evaluation, we introduce SAGE (Safety AI Generic Evaluation), an automated modular framework designed for customized and dynamic harm evaluations. SAGE employs prompted adversarial agents with diverse personalities based on the Big Five model, enabling system-aware multi-turn conversations that adapt to target applications and harm policies. We evaluate seven state-of-the-art LLMs across three applications and harm policies. Multi-turn experiments show that harm increases with conversation length, model behavior varies significantly when exposed to different user personalities and scenarios, and some models minimize harm via high refusal rates that reduce usefulness. We also demonstrate policy sensitivity within a harm category where tightening a child-focused sexual policy substantially increases measured defects across applications. These results motivate adaptive, policy-aware, and context-specific testing for safer real-world deployment.