Daily Papers

Current safety evaluations for LLM-driven agents primarily focus on atomic harms, failing to address sophisticated threats where malicious intent is concealed or diluted within complex tasks. We address this gap with a two-dimensional analysis of agent safety brittleness under the orthogonal pressures of intent concealment and task complexity. To enable this, we introduce OASIS (Orthogonal Agent Safety Inquiry Suite), a hierarchical benchmark with fine-grained annotations and a high-fidelity simulation sandbox. Our findings reveal two critical phenomena: safety alignment degrades sharply and predictably as intent becomes obscured, and a "Complexity Paradox" emerges, where agents seem safer on harder tasks only due to capability limitations. By releasing OASIS and its simulation environment, we provide a principled foundation for probing and strengthening agent safety in these overlooked dimensions.

Since the last nuclear atmospheric test carried out by the People Republic of China in 1980 and since the Chernobyl accident in 1986, the plutonium hasn't been directly released into the atmosphere. However, nowadays, it is still present in the troposphere. This is due to plutonium-bearing soil particles physical resuspension processes. In this work, we study for the first time the temporal variation of plutonium isotopes, 239Pu and 240Pu, baseline concentrations on a monthly basis in surface air from Seville (Spain), and their correlation with some tracers of mineral dust, during 2001 and 2002. The Pu analyses were performed by low-energy Accelerator Mass Spectrometry (AMS). The 239Pu plus 240Pu (239+240Pu) activity levels achieved maximums during the summer period, characterized by the absence of rains, and minimums during the rainy seasons, laying in the range 1-20 nBq per cubic meter. The 240Pu/239Pu two-year average atomic ratio was 0.18(0.03), in agreement with the fallout plutonium. A good correlation with Pu and Al and Ti levels is observed. They are crustal components usually used as tracers of African dust over European countries. The hypothesis of the influence of the Saharan dust intrusions is supported as well through the study of Total Ozone Mass Spectrometer (TOMS) daily images.

Understanding electron irradiation effects is vital not only for reliable transmission electron microscopy characterization, but increasingly also for the controlled manipulation of two-dimensional materials. The displacement cross sections of monolayer hBN are measured using aberration-corrected scanning transmission electron microscopy in near ultra-high vacuum at primary beam energies between 50 and 90 keV. Damage rates below 80 keV are up to three orders of magnitude lower than previously measured at edges under poorer residual vacuum conditions where chemical etching appears to have been dominant. Notably, is possible to create single vacancies in hBN using electron irradiation, with boron almost twice as likely as nitrogen to be ejected below 80 keV. Moreover, any damage at such low energies cannot be explained by elastic knock-on, even when accounting for vibrations of the atoms. A theoretical description is developed to account for lowering of the displacement threshold due to valence ionization resulting from inelastic scattering of probe electrons, modelled using charge-constrained density functional theory molecular dynamics. Although significant reductions are found depending on the constrained charge, quantitative predictions for realistic ionization states are currently not possible. Nonetheless, there is potential for defect-engineering of hBN at the level of single vacancies using electron irradiation.