Safety-Constrained Reinforcement Learning with Post-Training Reachability Verification for Robot Navigation

Safe navigation for mobile robots demands policies that remain reliable under the high-consequence perception uncertainty of cluttered environments. Yet most existing safe reinforcement learning (RL) methods assess safety through average cumulative cost. Such metrics can mask dangerous tail-risk behaviors. To address this, we propose a framework that trains risk-sensitive policies through Conditional Value-at-Risk (CVaR) constrained optimization on an off-policy TD3 backbone and evaluates their safety margins post-training through neural network reachability verification. During training, the policy is optimized under CVaR constraints on cumulative costs, promoting sensitivity to high-cost tail outcomes rather than average behavior alone. After training, we compute action reachable sets under bounded observation uncertainty using Taylor Model analysis, yielding a safety rate metric that quantifies the proportion of evaluated states at which the policy's reachable action set remains within prescribed safety margins. A key finding is that policies trained with CVaR constraints maintain larger safety margins from obstacles across evaluated states. This makes them significantly more amenable to formal reachability verification. Experiments across ten navigation scenarios and six baselines show that our method achieves a 98.3\% success rate, the highest safety verification rate among all compared methods, while revealing that average cost rankings and reachability-based safety rankings can diverge. This indicates that reachability verification captures risks which are missed by empirical cost metrics alone. We further validate our approach on a physical Clearpath Jackal robot, demonstrating successful sim-to-real transfer.

PrefixGuard: From LLM-Agent Traces to Online Failure-Warning Monitors

Large language model (LLM) agents now execute long, tool-using tasks where final outcome checks can arrive too late for intervention. Online warning requires lightweight prefix monitors over heterogeneous traces, but hand-authored event schemas are brittle and deployment-time LLM judging is costly. We introduce PrefixGuard, a trace-to-monitor framework with an offline StepView induction step followed by supervised monitor training. StepView induces deterministic typed-step adapters from raw trace samples, and the monitor learns an event abstraction and prefix-risk scorer from terminal outcomes. Across WebArena, $τ^2$-Bench, SkillsBench, and TerminalBench, the strongest PrefixGuard monitors reach 0.900/0.710/0.533/0.557 AUPRC. Using the strongest backend within each representation, they improve over raw-text controls by an average of +0.137 AUPRC. LLM judges remain substantially weaker under the same prefix-warning protocol. We also derive an observability ceiling on score-based area under the precision-recall curve (AUPRC) that separates monitor error from failures lacking evidence in the observed prefix. For finite-state audit, post-hoc deterministic finite automaton (DFA) extraction remains compact on WebArena and $τ^2$-Bench (29 and 20 states) but expands to 151 and 187 states on SkillsBench and TerminalBench. Finally, first-alert diagnostics show that strong ranking does not imply deployment utility: WebArena ranks well yet fails to support low-false-alarm alerts, whereas $τ^2$-Bench and TerminalBench retain more actionable early alerts. Together, these results position PrefixGuard as a practical monitor-synthesis recipe with explicit diagnostics for when prefix warnings translate into actionable interventions.

Lying with Truths: Open-Channel Multi-Agent Collusion for Belief Manipulation via Generative Montage

As large language models (LLMs) transition to autonomous agents synthesizing real-time information, their reasoning capabilities introduce an unexpected attack surface. This paper introduces a novel threat where colluding agents steer victim beliefs using only truthful evidence fragments distributed through public channels, without relying on covert communications, backdoors, or falsified documents. By exploiting LLMs' overthinking tendency, we formalize the first cognitive collusion attack and propose Generative Montage: a Writer-Editor-Director framework that constructs deceptive narratives through adversarial debate and coordinated posting of evidence fragments, causing victims to internalize and propagate fabricated conclusions. To study this risk, we develop CoPHEME, a dataset derived from real-world rumor events, and simulate attacks across diverse LLM families. Our results show pervasive vulnerability across 14 LLM families: attack success rates reach 74.4% for proprietary models and 70.6% for open-weights models. Counterintuitively, stronger reasoning capabilities increase susceptibility, with reasoning-specialized models showing higher attack success than base models or prompts. Furthermore, these false beliefs then cascade to downstream judges, achieving over 60% deception rates, highlighting a socio-technical vulnerability in how LLM-based agents interact with dynamic information environments. Our implementation and data are available at: https://github.com/CharlesJW222/Lying_with_Truth/tree/main.

Chasing Consistency: Quantifying and Optimizing Human-Model Alignment in Chain-of-Thought Reasoning

This paper presents a framework for evaluating and optimizing reasoning consistency in Large Language Models (LLMs) via a new metric, the Alignment Score, which quantifies the semantic alignment between model-generated reasoning chains and human-written reference chains in Chain-of-Thought (CoT) reasoning. Empirically, we find that 2-hop reasoning chains achieve the highest Alignment Score. To explain this phenomenon, we define four key error types: logical disconnection, thematic shift, redundant reasoning, and causal reversal, and show how each contributes to the degradation of the Alignment Score. Building on this analysis, we further propose Semantic Consistency Optimization Sampling (SCOS), a method that samples and favors chains with minimal alignment errors, significantly improving Alignment Scores by an average of 29.84% with longer reasoning chains, such as in 3-hop tasks.