Data to 14 Oct 2026

IS-Bench: Evaluating Interactive Safety of VLM-Driven Embodied Agents in Daily Household Tasks

arXiv 2506.16402 ↗First posted 19 Jun 2025cs.AI cs.CL cs.CVcited by 41

On the map: LLM agents and tool use → Agent threats, guardrails and governance

Abstract (arXiv)

Flawed planning from VLM-driven embodied agents poses significant safety hazards, hindering their deployment in real-world household tasks. However, existing static, non-interactive evaluation paradigms fail to adequately assess risks within these interactive environments, since they cannot simulate dynamic risks that emerge from an agent's actions and rely on unreliable post-hoc evaluations that ignore unsafe intermediate steps. To bridge this critical gap, we propose evaluating an agent's interactive safety: its ability to perceive emergent risks and execute mitigation steps in the correct procedural order. We thus present IS-Bench, the first multi-modal benchmark designed for interactive safety, featuring 161 challenging scenarios with 388 unique safety risks instantiated in a high-fidelity simulator. Crucially, it facilitates a novel process-oriented evaluation that verifies whether risk mitigation actions are performed before/after specific risk-prone steps. Extensive experiments on leading VLMs, including the GPT-4o and Gemini-2.5 series, reveal that current agents lack interactive safety awareness, and that while safety-aware Chain-of-Thought can improve performance, it often compromises task completion. By highlighting these critical limitations, IS-Bench provides a foundation for developing safer and more reliable embodied AI systems. Code and data are released under https://github.com/AI45Lab/IS-Bench.

Key facts. Cited by 41 papers in the map (data through 14 Oct 2026). Placed in Agent threats, guardrails and governance, one of 16 directions in LLM agents and tool use. Builds on 34 of its 46 references that are in the map. First posted on arXiv 19 Jun 2025.

Builds on

34 of 46 references.

Closest ancestors · 5

Same direction, newest first

Other references · 29

2022ReAct: Synergizing Reasoning and Acting in Language Modelscited by 8,695
2025Qwen2.5-VL Technical Reportcited by 5,587
2024GPT-4o System Cardcited by 4,588
2023PaLM-E: An Embodied Multimodal Language Modelcited by 2,071
2022Language Models as Zero-Shot Planners: Extracting Actionable Knowledge for Embodied Agentscited by 1,035
2022ProgPrompt: Generating Situated Robot Task Plans using Large Language Modelscited by 611
2023DecodingTrust: A Comprehensive Assessment of Trustworthiness in GPT Modelscited by 518
2022LLM-Planner: Few-Shot Grounded Planning for Embodied Agents with Large Language Modelscited by 462
2022Learning to Prompt for Open-Vocabulary Object Detection with Vision-Language Modelcited by 317
2023SayPlan: Grounding Large Language Models using 3D Scene Graphs for Scalable Robot Task Planningcited by 294
2023NavGPT: Explicit Reasoning in Vision-and-Language Navigation with Large Language Modelscited by 280
2023EmbodiedGPT: Vision-Language Pre-Training via Embodied Chain of Thoughtcited by 279

17 more

2023MM-SafetyBench: A Benchmark for Safety Evaluation of Multimodal Large Language Modelscited by 274
2021A Survey of Embodied AI: From Simulators to Research Taskscited by 261
2022Partially Observable Markov Decision Processes in Robotics: A Surveycited by 86
2024VLSBench: Unveiling Visual Leakage in Multimodal Safetycited by 63
2024Multimodal Situational Safetycited by 58
2024A Survey on Robotics with Foundation Models: toward Embodied AIcited by 56
2023Paxion: Patching Action Knowledge in Video-Language Foundation Modelscited by 42
2025Towards Robust and Secure Embodied AI: A Survey on Vulnerabilities and Attackscited by 34
2024Exploring the Robustness of Decision-Level Through Adversarial Attacks on LLM-Based Embodied Modelscited by 33
2023Octavius: Mitigating Task Interference in MLLMs via LoRA-MoEcited by 25
2024RH20T-P: A Primitive-Level Robotic Dataset Towards Composable Generalization Agentscited by 20
2025X-Boundary: Establishing Exact Safety Boundary to Shield LLMs from Multi-Turn Jailbreaks without Compromising Usabilitycited by 20
2025Subtle Risks, Critical Failures: A Framework for Diagnosing Physical Safety of LLMs for Embodied Decision Makingcited by 20
2024MLDT: Multi-Level Decomposition for Complex Long-Horizon Robotic Task Planning with Open-Source Large Language Modelcited by 15
2025RoboDexVLM: Visual Language Model-Enabled Task Planning and Motion Control for Dexterous Robot Manipulationcited by 15
2023ViStruct: Visual Structural Knowledge Extraction via Curriculum Guided Code-Vision Representationcited by 13
2024Safe Planner: Empowering Safety Awareness in Large Pre-Trained Models for Robot Task Planningcited by 9

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 6; 2026: 35.

41 papers.

Sort:

By area:

2025Geometrically-Constrained Agent for Spatial Reasoningcited by 26
2025A Comprehensive Survey on Trustworthiness in Reasoning with Large Language Modelscited by 25
2025SafeMind: Benchmarking and Mitigating Safety Risks in Embodied LLM Agentscited by 17
2026HazardArena: Evaluating Semantic Safety in Vision-Language-Action Modelscited by 8
2025MADRA: Multi-Agent Debate for Risk-Aware Embodied Planningcited by 6
2025HomeSafeBench: Benchmarking Embodied Vision-Language Models in Free-Exploration Home Safety Inspectioncited by 5
2026HomeGuard: VLM-based Embodied Safeguard for Identifying Contextual Risk in Household Taskcited by 5
2026Using large language models for embodied planning introduces systematic safety riskscited by 5
2026SAVeS: Steering Safety Judgments in Vision-Language Models via Semantic Cuescited by 4
2026BeSafe-Bench: Unveiling Behavioral Safety Risks of Situated Agents in Functional Environmentscited by 4
2026Taxonomy and Consistency Analysis of Safety Benchmarks for AI Agentscited by 4
2026Concept-Based Dictionary Learning for Inference-Time Safety in Vision Language Action Modelscited by 3

29 more

Similar, not cited

Close in wording, no citation either way