When AI interacts with the physical world -- as a robot or an assistive agent -- new safety challenges emerge beyond those of purely ``digital AI". In such interactions, the potential for physical harm is direct and immediate. How well do state-of-the-art foundation models understand common-sense facts about physical safety, e.g. that a box may be too heavy to lift, or that a hot cup of coffee should not be handed to a child? In this paper, our contributions are three-fold: first, we develop a highly scalable approach to continuous physical safety benchmarking of Embodied AI systems, grounded in real-world injury narratives and operational safety constraints. To probe multi-modal safety understanding, we turn these narratives and constraints into photorealistic images and videos capturing transitions from safe to unsafe states, using advanced generative models. Secondly, we comprehensively analyze the ability of major foundation models to perceive risks, reason about safety, and trigger interventions; this yields multi-faceted insights into their deployment readiness for safety-critical agentic applications. Finally, we develop a post-training paradigm to teach models to explicitly reason about embodiment-specific safety constraints provided through system instructions. The resulting models generate thinking traces that make safety reasoning interpretable and transparent, achieving state of the art performance in constraint satisfaction evaluations. The benchmark is released at https://asimov-benchmark.github.io/v2

Large Language Models (LLMs) are increasingly used for decision making in embodied agents, yet existing safety evaluations often rely on coarse success rates and domain-specific setups, making it difficult to diagnose why and where these models fail. This obscures our understanding of embodied safety and limits the selective deployment of LLMs in high-risk physical environments. We introduce SAFEL, the framework for systematically evaluating the physical safety of LLMs in embodied decision making. SAFEL assesses two key competencies: (1) rejecting unsafe commands via the Command Refusal Test, and (2) generating safe and executable plans via the Plan Safety Test. Critically, the latter is decomposed into functional modules, goal interpretation, transition modeling, action sequencing, enabling fine-grained diagnosis of safety failures. To support this framework, we introduce EMBODYGUARD, a PDDL-grounded benchmark containing 942 LLM-generated scenarios covering both overtly malicious and contextually hazardous instructions. Evaluation across 13 state-of-the-art LLMs reveals that while models often reject clearly unsafe commands, they struggle to anticipate and mitigate subtle, situational risks. Our results highlight critical limitations in current LLMs and provide a foundation for more targeted, modular improvements in safe embodied reasoning.

Large Language Models (LLMs) are increasingly used to control robotic systems such as drones, but their risks of causing physical threats and harm in real-world applications remain unexplored. Our study addresses the critical gap in evaluating LLM physical safety by developing a comprehensive benchmark for drone control. We classify the physical safety risks of drones into four categories: (1) human-targeted threats, (2) object-targeted threats, (3) infrastructure attacks, and (4) regulatory violations. Our evaluation of mainstream LLMs reveals an undesirable trade-off between utility and safety, with models that excel in code generation often performing poorly in crucial safety aspects. Furthermore, while incorporating advanced prompt engineering techniques such as In-Context Learning and Chain-of-Thought can improve safety, these methods still struggle to identify unintentional attacks. In addition, larger models demonstrate better safety capabilities, particularly in refusing dangerous commands. Our findings and benchmark can facilitate the design and evaluation of physical safety for LLMs.

The primary aim of this paper is to investigate the integration of emotions into the social navigation framework to analyse its effect on both navigation and human physiological safety and comfort. The proposed framework uses leg detection to find the whereabouts of people and computes adaptive proxemic zones based on their emotional state. We designed several case studies in a simulated environment and examined 3 different emotions; positive (happy), neutral and negative (angry). A survey study was conducted with 70 participants to explore their impressions about the navigation of the robot and compare the human safety and comfort measurements results. Both survey and simulation results showed that integrating emotions into proxemic zones has a significant effect on the physical safety of a human. The results revealed that when a person is angry, the robot is expected to navigate further than the standard distance to support his/her physiological comfort and safety. The results also showed that reducing the navigation distance is not preferred when a person is happy.

Understanding what constitutes safe text is an important issue in natural language processing and can often prevent the deployment of models deemed harmful and unsafe. One such type of safety that has been scarcely studied is commonsense physical safety, i.e. text that is not explicitly violent and requires additional commonsense knowledge to comprehend that it leads to physical harm. We create the first benchmark dataset, SafeText, comprising real-life scenarios with paired safe and physically unsafe pieces of advice. We utilize SafeText to empirically study commonsense physical safety across various models designed for text generation and commonsense reasoning tasks. We find that state-of-the-art large language models are susceptible to the generation of unsafe text and have difficulty rejecting unsafe advice. As a result, we argue for further studies of safety and the assessment of commonsense physical safety in models before release.

With the powerful performance of Artificial Intelligence (AI) also comes prevalent ethical issues. Though governments and corporations have curated multiple AI ethics guidelines to curb unethical behavior of AI, the effect has been limited, probably due to the vagueness of the guidelines. In this paper, we take a closer look at how AI ethics issues take place in real world, in order to have a more in-depth and nuanced understanding of different ethical issues as well as their social impact. With a content analysis of AI Incident Database, which is an effort to prevent repeated real world AI failures by cataloging incidents, we identified 13 application areas which often see unethical use of AI, with intelligent service robots, language/vision models and autonomous driving taking the lead. Ethical issues appear in 8 different forms, from inappropriate use and racial discrimination, to physical safety and unfair algorithm. With this taxonomy of AI ethics issues, we aim to provide AI practitioners with a practical guideline when trying to deploy AI applications ethically.

Tina D'Agostin is CEO of Alcatraz AI, a leader in physical security technologies using facial authentication for autonomous access control. After a pandemic year, workplace safety is top of mind for many businesses. According to one survey, employee safety and well-being replaced pre-existing priorities, like attracting top talent, as the dominant concerns for business leaders. Meanwhile, many employees are dissatisfied with existing safety measures. A Gallup Poll found that just 65% of workers are "completely satisfied with their physical safety at work," the lowest point since 2001.

Tina D'Agostin is CEO of Alcatraz AI, a leader in physical security technologies using facial authentication for autonomous access control. After a pandemic year, workplace safety is top of mind for many businesses. According to one survey, employee safety and well-being replaced pre-existing priorities, like attracting top talent, as the dominant concerns for business leaders. Meanwhile, many employees are dissatisfied with existing safety measures. A Gallup Poll found that just 65% of workers are "completely satisfied with their physical safety at work," the lowest point since 2001.

It's hard to believe, but a year in which the unprecedented seemed to happen every day is just weeks from being over. In AI circles, the end of the calendar year means the rollout of annual reports aimed at defining progress, impact, and areas for improvement. The AI Index is due out in the coming weeks, as is CB Insights' assessment of global AI startup activity, but two reports -- both called The State of AI -- have already been released. Last week, McKinsey released its global survey on the state of AI, a report now in its third year. Interviews with executives and a survey of business respondents found a potential widening of the gap between businesses that apply AI and those that do not.