Searcharxiv⌕ Search

arXiv subjects

Charles Xiao

Publications and source records attributed to Charles Xiao.

6 recordsLinked to original sources

Safe Resemblance, Silent Propagation: Limits of Text-Similarity Guardrails for AI Agent Safety

When AI agents relay information to one another, an output can preserve an injected instruction's effect on downstream agents even after its wording changes, yet legitimate quotation and template reuse can produce high textual overlap with the source, so text-similarity guardrails that attempt to flag the former risk conflating these two cases. We test this distinction using DonkeyRail, a public classifier combining BLEU, ROUGE-L, and METEOR features, deployed in a controlled three-agent email relay, defining a behavioral measure in which an output counts as functional (the injected procedure induces fresh computations in two subsequent agents) independent of the detector's score, alongside legitimate controls that explicitly authorize quotation and reuse. At its default threshold the classifier misses 8 of 19 functional outputs while flagging 70 of 192 legitimate replies, with every completed quotation and template-reuse reply triggering a false alert; calibrating for low false positives on one template set does not transfer, as the recalibrated threshold detects only one functional output and still produces 15 false alerts on new templates. We also identify an operating-point error in all 21 published classifier configurations, where ROC interpolation replaces direct threshold application, and show formally that high AUC against a surface-level text criterion provides no distribution-free guarantee of high AUC against the behavioral target, while the existing evaluation labels lack the joint observations needed to estimate the gap. An authorization analysis further shows that identical visible transcripts can require different safety decisions depending on the governing policy, a distinction no observation-only classifier can resolve. These results argue for behavioral and policy-grounded evaluation over text-similarity guardrails.

cs.AI↗

When Safety Becomes an Attack Surface: Manipulating Abstention in Tool-Using AI Agents

An AI agent's decision not to act can itself become an attack surface. We study warning-induced over-abstention, an availability attack in which adversaries place fabricated risk warnings in low-trust content to prevent legitimate task completion. Our central premise is that safe abstention requires evaluating both the claimed risk and the authority of the evidence supporting it. We develop controlled counterfactual task variants that vary risk content, source provenance, and warning wording, and evaluate a Verify-Before-Abstain protocol that corroborates warnings against authoritative records or the underlying content. Across three model families and 51 injectable AgentAbstain task pairs, caution-first prompting amplifies the attack. For Gemini, induced over-abstention rises from $9.8\%$ to $86.3\%$, despite unchanged clean-task completion. Matched wording comparisons further reveal sensitivity to warning language, with substantial variation across models and environments. Provenance-aware verification improves legitimate completion, while forgery and sensitive-content controls expose its dependence on authoritative evidence and its limitations in preserving confidentiality. These results motivate evaluating agent safety through the justification for restraint, alongside the safety of actions taken.

cs.CR↗

Passively Adaptive Radiative Switch for Thermoregulation in Buildings

With the ever-growing need to reduce energy consumption, building materials that passively heat or cool are gaining importance. However, many buildings require both heating and cooling, even within the same day. To date, few technologies can automatically switch between passive heating and cooling, and those that can require a large temperature range to cycle states (>15o C), making them ineffective for daily switching. We present a passively adaptive radiative switch that leverages the expansion in phase-change energy storage materials to actuate the motion of louvers and can cycle states in less than 3o C. The black selective-absorber louvers induce high heat gain when closed, yet when open, expose a white, emissive surface for low heat gain. During an outdoor test in which temperature was held steady, our device reduced the energetic cost of cooling by 3.1x and heating by 2.6x compared to non-switching devices. Our concept opens the door for passively adaptive thermoregulating building materials.

physics.app-ph↗

A light- and heat-seeking vine-inspired robot with material-level responsiveness

The fields of soft and bio-inspired robotics promise to imbue synthetic systems with capabilities found in the natural world. However, many of these biological capabilities are yet to be realized. For example, vines in nature direct growth via localized responses embedded in the cells of vine body, allowing an organism without a central brain to successfully search for resources (e.g., light). Yet to date, vine-inspired robots have yet to show such localized embedded responsiveness. Here we present a vine-inspired robotic device with material-level responses embedded in its skin and capable of growing and steering toward either a light or heat stimulus. We present basic modeling of the concept, design details, and experimental results showing its behavior in response to infrared (IR) and visible light. Our simple design concept advances the capabilities of bio-inspired robots and lays the foundation for future growing robots that are capable of seeking light or heat, yet are extremely simple and low-cost. Potential applications include solar tracking, and in the future, firefighting smoldering fires. We envision using similar robots to find hot spots in hard-to-access environments, allowing us to put out potentially long-burning fires faster.

cs.RO↗

Field-mediated locomotor dynamics on highly deformable surfaces

In many systems motion occurs on deformed and deformable surfaces, setting up the possibility for dynamical interactions solely mediated by the coupling of the entities with their environment. Here we study the "two-body" dynamics of robot locomotion on a highly deformable spandex membrane in two scenarios: one in which a robot orbits a large central depression and the other where the two robots affect each other's motion solely through mutual environmental deformations. Inspired by the resemblance of the orbits of the single robot with those of general relativistic orbits around black holes, we recast the vehicle plus membrane dynamics in physical space into the geodesic motion of a "test particle" in a fiducial curved space-time and demonstrate how this framework facilitates understanding the observed dynamics. The two-robot problem also exhibits a resemblance with Einstein's general relativistic view of gravity, which in the words of Wheeler: "spacetime tells matter how to move; matter tells spacetime how to curve." We generalize this case the mapping to include a reciprocal coupling that translates into robotic curvature-based control schemes which modify interaction (promoting avoidance or aggregation) without long-range sensing. Our work provides a starting point for developing a mechanical analog gravity system as well as develops a framework that can provide insights into active matter in deformable environments and robot exploration in complex landscapes.

cs.RO↗

Thermodynamically-informed Air-based Soft Heat Engine Design

Soft heat engines are poised to play a vital role in future soft robots due to their easy integration into soft structures and low-voltage power requirements. Recent works have demonstrated soft heat engines relying on liquid-to-gas phase change materials. However, despite the fact that many soft robots have air as a primary component, soft air cycles are not a focus of the field. In this paper, we develop theory for air-based soft heat engines design and efficiency, and demonstrate experimentally that efficiency can be improved through careful cycle design. We compare a simple constant-load cycle to a designed decreasing-load cycle, inspired by the Otto cycle. While both efficiencies are relatively low, the Otto-like cycle improves efficiency by a factor of 11.3, demonstrating the promise of this approach. Our results lay the foundation for the development of air-based soft heat engines as a new option for powering soft robots.

cs.RO↗