We introduce a unified mathematical and probabilistic framework for understanding and comparing diverse AI agent strategies. We bridge the gap between high-level agent design concepts, such as ReAct, multi-agent systems, and control flows, and a rigorous mathematical formulation. Our approach frames agentic processes as a chain of probabilities, enabling a detailed analysis of how different strategies manipulate these probabilities to achieve desired outcomes. Our framework provides a common language for discussing the trade-offs inherent in various agent architectures. One of our many key contributions is the introduction of the "Degrees of Freedom" concept, which intuitively differentiates the optimizable levers available for each approach, thereby guiding the selection of appropriate strategies for specific tasks. This work aims to enhance the clarity and precision in designing and evaluating AI agents, offering insights into maximizing the probability of successful actions within complex agentic systems.

To collaborate with humans, robots must infer goals that are often ambiguous, difficult to articulate, or not drawn from a fixed set. Prior approaches restrict inference to a predefined goal set, rely only on observed actions, or depend exclusively on explicit instructions, making them brittle in real-world interactions. We present BALI (Bidirectional Action-Language Inference) for goal prediction, a method that integrates natural language preferences with observed human actions in a receding-horizon planning tree. BALI combines language and action cues from the human, asks clarifying questions only when the expected information gain from the answer outweighs the cost of interruption, and selects supportive actions that align with inferred goals. We evaluate the approach in collaborative cooking tasks, where goals may be novel to the robot and unbounded. Compared to baselines, BALI yields more stable goal predictions and significantly fewer mistakes.

Automating disassembly of critical components from end-of-life (EoL) desktops, such as high-value items like RAM modules and CPUs, as well as sensitive parts like hard disk drives, remains challenging due to the inherent variability and uncertainty of these products. Moreover, their disassembly requires sequential, precise, and dexterous operations, further increasing the complexity of automation. Current robotic disassembly processes are typically divided into several stages: perception, sequence planning, task planning, motion planning, and manipulation. Each stage requires explicit modeling, which limits generalization to unfamiliar scenarios. Recent development of vision-language-action (VLA) models has presented an end-to-end approach for general robotic manipulation tasks. Although VLAs have demonstrated promising performance on simple tasks, the feasibility of applying such models to complex disassembly remains largely unexplored. In this paper, we collected a customized dataset for robotic RAM and CPU disassembly and used it to fine-tune two well-established VLA approaches, OpenVLA and OpenVLA-OFT, as a case study. We divided the whole disassembly task into several small steps, and our preliminary experimental results indicate that the fine-tuned VLA models can faithfully complete multiple early steps but struggle with certain critical subtasks, leading to task failure. However, we observed that a simple hybrid strategy that combines VLA with a rule-based controller can successfully perform the entire disassembly operation. These findings highlight the current limitations of VLA models in handling the dexterity and precision required for robotic EoL product disassembly. By offering a detailed analysis of the observed results, this study provides insights that may inform future research to address current challenges and advance end-to-end robotic automated disassembly.

The evolution toward sixth-generation wireless systems positions intelligence as a native network capability, fundamentally transforming the design of radio access networks (RANs). Within this vision, Semantic-native communication and agentic intelligence are expected to play central roles. SemCom departs from bit-level fidelity and instead emphasizes task-oriented meaning exchange, enabling compact SC and introducing new performance measures such as semantic fidelity and task success rate. Agentic intelligence endows distributed RAN entities with goal-driven autonomy, reasoning, planning, and multi-agent collaboration, increasingly supported by foundation models and knowledge graphs. In this work, we first introduce the conceptual foundations of SemCom and agentic networking, and discuss why existing AI-driven O-RAN solutions remain largely bit-centric and task-siloed. We then present a unified taxonomy that organizes recent research along three axes: i) semantic abstraction level (symbol/feature/intent/knowledge), ii) agent autonomy and coordination granularity (single-, multi-, and hierarchical-agent), and iii) RAN control placement across PHY/MAC, near-real-time RIC, and non-real-time RIC. Based on this taxonomy, we systematically introduce enabling technologies including task-oriented semantic encoders/decoders, multi-agent reinforcement learning, foundation-model-assisted RAN agents, and knowledge-graph-based reasoning for cross-layer awareness. Representative 6G use cases, such as immersive XR, vehicular V2X, and industrial digital twins, are analyzed to illustrate the semantic-agentic convergence in practice. Finally, we identify open challenges in semantic representation standardization, scalable trustworthy agent coordination, O-RAN interoperability, and energy-efficient AI deployment, and outline research directions toward operational semantic-agentic AI-RAN.

The rapid advancement of Large Language Models (LLMs) is catalyzing a shift towards autonomous AI Agents capable of executing complex, multi-step tasks. However, these agents remain brittle when faced with real-world exceptions, making Human-in-the-Loop (HITL) supervision essential for mission-critical applications. In this paper, we present AgentBay, a novel sandbox service designed from the ground up for hybrid interaction. AgentBay provides secure, isolated execution environments spanning Windows, Linux, Android, Web Browsers, and Code interpreters. Its core contribution is a unified session accessible via a hybrid control interface: An AI agent can interact programmatically via mainstream interfaces (MCP, Open Source SDK), while a human operator can, at any moment, seamlessly take over full manual control. This seamless intervention is enabled by Adaptive Streaming Protocol (ASP). Unlike traditional VNC/RDP, ASP is specifically engineered for this hybrid use case, delivering an ultra-low-latency, smoother user experience that remains resilient even in weak network environments. It achieves this by dynamically blending command-based and video-based streaming, adapting its encoding strategy based on network conditions and the current controller (AI or human). Our evaluation demonstrates strong results in security, performance, and task completion rates. In a benchmark of complex tasks, the AgentBay (Agent + Human) model achieved more than 48% success rate improvement. Furthermore, our ASP protocol reduces bandwidth consumption by up to 50% compared to standard RDP, and in end-to-end latency with around 5% reduction, especially under poor network conditions. We posit that AgentBay provides a foundational primitive for building the next generation of reliable, human-supervised autonomous systems.

Recent advances in large multimodal models have enabled new opportunities in embodied AI, particularly in robotic manipulation. These models have shown strong potential in generalization and reasoning, but achieving reliable and responsible robotic behavior in real-world settings remains an open challenge. In high-stakes environments, robotic agents must go beyond basic task execution to perform risk-aware reasoning, moral decision-making, and physically grounded planning. We introduce ResponsibleRobotBench, a systematic benchmark designed to evaluate and accelerate progress in responsible robotic manipulation from simulation to real world. This benchmark consists of 23 multi-stage tasks spanning diverse risk types, including electrical, chemical, and human-related hazards, and varying levels of physical and planning complexity. These tasks require agents to detect and mitigate risks, reason about safety, plan sequences of actions, and engage human assistance when necessary. Our benchmark includes a general-purpose evaluation framework that supports multimodal model-based agents with various action representation modalities. The framework integrates visual perception, context learning, prompt construction, hazard detection, reasoning and planning, and physical execution. It also provides a rich multimodal dataset, supports reproducible experiments, and includes standardized metrics such as success rate, safety rate, and safe success rate. Through extensive experimental setups, ResponsibleRobotBench enables analysis across risk categories, task types, and agent configurations. By emphasizing physical reliability, generalization, and safety in decision-making, this benchmark provides a foundation for advancing the development of trustworthy, real-world responsible dexterous robotic systems. https://sites.google.com/view/responsible-robotbench

Benchmarks are the primary tool for assessing progress in artificial intelligence (AI), yet current practice evaluates models on isolated test suites and provides little guidance for reasoning about generality or autonomous self-improvement. Here we introduce a geometric framework in which all psychometric batteries for AI agents are treated as points in a structured moduli space, and agent performance is described by capability functionals over this space. First, we define an Autonomous AI (AAI) Scale, a Kardashev-style hierarchy of autonomy grounded in measurable performance on batteries spanning families of tasks (for example reasoning, planning, tool use and long-horizon control). Second, we construct a moduli space of batteries, identifying equivalence classes of benchmarks that are indistinguishable at the level of agent orderings and capability inferences. This geometry yields determinacy results: dense families of batteries suffice to certify performance on entire regions of task space. Third, we introduce a general Generator-Verifier-Updater (GVU) operator that subsumes reinforcement learning, self-play, debate and verifier-based fine-tuning as special cases, and we define a self-improvement coefficient $κ$ as the Lie derivative of a capability functional along the induced flow. A variance inequality on the combined noise of generation and verification provides sufficient conditions for $κ> 0$. Our results suggest that progress toward artificial general intelligence (AGI) is best understood as a flow on moduli of benchmarks, driven by GVU dynamics rather than by scores on individual leaderboards.

Multimodal Large Language Models (MLLMs) show promising results as decision-making engines for embodied agents operating in complex, physical environments. However, existing benchmarks often prioritize high-level planning or spatial reasoning, leaving the fine-grained action intelligence required for embodied physical interaction un-derexplored. T o address this gap, we introduce CFG-Bench, a new benchmark designed to systematically evaluate this crucial capability. CFG-Bench consists of 1,368 curated videos paired with 19,562 three-modalities question-answer pairs targeting four cognitive abilities: 1) Physical Interaction, 2) T emporal-Causal Relation, 3) Intentional Understanding, and 4) Evaluative Judgment. T ogether, these dimensions provide a systematic framework for assessing a model's ability to translate visual observations into actionable knowledge, moving beyond mere surface-level recognition. Our comprehensive evaluation on CFG-Bench reveals that leading MLLMs struggle to produce detailed instructions for physical interactions and exhibit profound limitations in the higher-order reasoning of intention and evaluation. Moreover, supervised fine-tuning (SFT) on our data demonstrates that teaching an MLLMs to articulate fine-grained actions directly translates to significant performance gains on established embodied benchmarks. Our analysis highlights these limitations and offers insights for developing more capable and grounded embodied agents.

LLMs generate buggy code: 29.6% of SWE-bench solved patches fail, 62% of BaxBench solutions have vulnerabilities, and existing tools only catch 65% of bugs with 35% false positives. We built CodeX-Verify, a multi-agent system that uses four specialized agents to detect different types of bugs. We prove mathematically that combining agents with different detection patterns finds more bugs than any single agent when the agents look for different problems, using submodularity of mutual information under conditional independence. Measuring agent correlation of rho = 0.05 to 0.25 confirms they detect different bugs. Testing on 99 code samples with verified labels shows our system catches 76.1% of bugs, matching the best existing method (Meta Prompt Testing: 75%) while running faster and without test execution. We tested all 15 agent combinations and found that using multiple agents improves accuracy by 39.7 percentage points (from 32.8% to 72.4%) compared to single agents, with diminishing returns of +14.9pp, +13.5pp, and +11.2pp for agents 2, 3, and 4, validating our theoretical model. The best two-agent combination (Correctness + Performance) reaches 79.3% accuracy. Testing on 300 real patches from Claude Sonnet 4.5 runs in under 200ms per sample, making this practical for production use.

AI agents can autonomously perform tasks and, often without explicit user consent, collect or disclose users' sensitive local data, which raises serious privacy concerns. Although AI agents' privacy policies describe their intended data practices, there remains limited transparency and accountability about whether runtime behavior matches those policies. To close this gap, we introduce AudAgent, a visual tool that continuously monitors AI agents' data practices in real time and guards compliance with stated privacy policies. AudAgent consists of four components for automated privacy auditing of AI agents. (i) Policy formalization: a novel cross-LLM voting mechanism to guarantee confidence of the parsed privacy policy model. (ii) Runtime annotation: a lightweight Presidio-based analyzer detects sensitive data and annotates data practices based on the AI agent's context and the privacy policy model. (iii) Compliance auditing: ontology graphs and automata-based checking connect the privacy policy model with runtime annotations, enabling on-the-fly compliance checking. (iv) User interface: an infrastructure-independent implementation visualizes the real-time execution trace of AI agents along with potential privacy policy violations, providing user-friendly transparency and accountability. We evaluate AudAgent with AI agents built using mainstream frameworks, demonstrating its effectiveness in detecting and visualizing privacy policy violations in real time. Using AudAgent, we also find that most privacy policies omit explicit safeguards for highly sensitive data such as SSNs, whose misuse violates legal requirements, and that many agents do not refuse handling such data via third-party tools, including those controlled by Claude, Gemini, and DeepSeek. AudAgent proactively blocks operations on such data, overriding the agents' original privacy policy and behavior.