Challenge: Current safety evaluation methodologies focus on single-turn interactions with generic policies, failing to capture conversational dynamics of real-world usage and application-specific harms.
Approach: They propose a framework for customized and dynamic harm evaluations that employs prompted adversarial agents with diverse personalities based on the Big Five model.
Outcome: The proposed framework enables system-aware multi-turn conversations that adapt to target applications and harm policies.

Similar Papers

MHSafeEval: Role-Aware Interaction-Level Evaluation of Mental Health Safety in Large Language Models (2026.findings-acl)

Copied to clipboard

Challenge: Existing evaluation frameworks assess isolated responses using coarse-grained taxonomies or static datasets.
Approach: They propose a role-aware mental health safety taxonomy that characterizes clinically significant harm in terms of interactional roles an AI counselor adopts.
Outcome: The proposed framework significantly improves failure-mode coverage and diagnostic granularity.
Dynamic Evaluation with Cognitive Reasoning for Multi-turn Safety of Large Language Models (2025.acl-long)

Copied to clipboard

Challenge: Existing safety evaluation methods rely on static assessments that use fixed harmful prompts or predefined prefixes as jailbreak templates.
Approach: They propose a dynamic evaluation framework for multi-turn safety assessment of LLMs based on cognitive theories to simulate real chatting process and scenario simulation and strategy decision to guide dynamic generation.
Outcome: The proposed framework has been applied to evaluate the safety of widely used LLMs.
AgenticEval: Toward Agentic and Self-Evolving Safety Evaluation of Large Language Models (2026.findings-acl)

Copied to clipboard

Challenge: Existing static safety evaluation methods are ill-equipped to address dynamic nature of AI risks and evolving regulations, creating a critical safety gap.
Approach: They propose a new paradigm of agentic safety evaluation reframing evaluation as a continuous and self-evolving process rather than a one-time audit.
Outcome: The proposed framework shows a consistent decline in model safety as the evaluation hardens.
Inverting the Shield: Systematically Generating Safety Tests from Policy Specifications (2026.acl-long)

Copied to clipboard

Challenge: Existing safety evaluation paradigms rely on constructed benchmarks or dynamic red-teaming to probe potential vulnerabilities.
Approach: They propose a framework that integrates specification-based software testing with AI safety.
Outcome: The proposed framework achieves higher coverage and attack success counts compared to baselines.
Persona-Grounded Safety Evaluation of AI Companions in Multi-Turn Conversations (2026.acl-long)

Copied to clipboard

Challenge: Existing safety evaluations rely on self-reported user data or interviews . a recent study evaluated how Replika responds to high-risk user groups .
Approach: They propose a framework for controlled simulation and safety evaluation of multi-turn interactions with AI companion applications.
Outcome: The proposed framework evaluates how Replika responds to high-risk user groups . it incorporates emotion modeling and LLM-assisted utterance-and harm-level classification .
DialogGuard: Multi-Agent Psychosocial Safety Evaluation Interface of Sensitive LLM Responses (2026.acl-demo)

Copied to clipboard

Challenge: Existing tools do not surface subtler psychosocial harms, nor provide explainable rationales that practitioners need.
Approach: They propose an open-source system that lets practitioners inspect, stress-test, and create audit trails for prompted LLM agents across five psychosocial safety dimensions.
Outcome: The open-source DialogGuard system lets practitioners inspect, stress-test, and create audit trails for prompted LLM agents across five psychosocial safety dimensions.
SafetyQuizzer: Timely and Dynamic Evaluation on the Safety of LLMs (2025.naacl-long)

Copied to clipboard

Challenge: Large Language Models (LLMs) have been used to evaluate the safety of their users . however, evaluation questions in current benchmarks are too straightforward and difficult to update with practical relevance due to their lack of correlation with real-world events.
Approach: They propose a question-generation framework to evaluate the safety of LLMs in the Chinese context.
Outcome: The proposed framework reduces decline rate while maintaining similar attack success rate.
From Representational Harms to Quality-of-Service Harms: A Case Study on Llama 2 Safety Safeguards (2024.findings-acl)

Copied to clipboard

Challenge: Recent advances in large language models have also introduced additional safety risks and raised concerns regarding their detrimental impact on already marginalized populations.
Approach: They propose to use LLMs to evaluate their safety responses on already mitigated biases by evaluating models on already encoded assumptions.
Outcome: The proposed model can encode harmful assumptions, but it can also be harmful for certain demographic groups.
Do-Not-Answer: Evaluating Safeguards in LLMs (2024.findings-eacl)

Copied to clipboard

Challenge: a dataset evaluating harmful capabilities in large language models is available at https://github.com/Libr-AI/do-not-answer.
Approach: They collect an open-source dataset to evaluate the safeguards in large language models . they find that simple BERT-style classifiers can achieve results comparable to GPT-4 .
Outcome: The proposed dataset compares the safety of six popular LLMs to GPT-4 on automatic safety evaluation.
TAMAS: Benchmarking Adversarial Risks in Multi-Agent LLM Systems (2026.acl-long)

Copied to clipboard

Challenge: Existing benchmarks and datasets focus on single-agent settings, failing to capture the unique vulnerabilities of multi-agend LLM dynamics and co-ordination.
Approach: They propose a benchmark to evaluate the robustness and safety of multi-agent LLM systems.
Outcome: The proposed benchmark evaluates the robustness and safety of multi-agent LLM systems.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations