The rapid deployment of large language model (LLM)-based agents introduces a new class of risks, driven by their capacity for autonomous planning, multi-step tool integration, and emergent interactions. It raises some risk factors for existing governance approaches as they remain fragmented: Existing frameworks are either static taxonomies driven; however, they lack an integrated end-to-end pipeline from risk identification to operational assurance, especially for an agentic platform. We propose AGENTSAFE, a practical governance framework for LLM-based agentic systems. The framework operationalises the AI Risk Repository into design, runtime, and audit controls, offering a governance framework for risk identification and assurance. The proposed framework, AGENTSAFE, profiles agentic loops (plan -> act -> observe -> reflect) and toolchains, and maps risks onto structured taxonomies extended with agent-specific vulnerabilities. It introduces safeguards that constrain risky behaviours, escalates high-impact actions to human oversight, and evaluates systems through pre-deployment scenario banks spanning security, privacy, fairness, and systemic safety. During deployment, AGENTSAFE ensures continuous governance through semantic telemetry, dynamic authorization, anomaly detection, and interruptibility mechanisms. Provenance and accountability are reinforced through cryptographic tracing and organizational controls, enabling measurable, auditable assurance across the lifecycle of agentic AI systems. The key contributions of this paper are: (1) a unified governance framework that translates risk taxonomies into actionable design, runtime, and audit controls; (2) an Agent Safety Evaluation methodology that provides measurable pre-deployment assurance; and (3) a set of runtime governance and accountability mechanisms that institutionalise trust in agentic AI ecosystems.

Physical spaces are increasingly dense with networked devices, promising seamless coordination and ambient intelligence. Yet today, cloud-first architectures force all communication through wide-area networks regardless of physical proximity. We lack an abstraction for spatial networking: using physical spaces to create boundaries for private, robust, and low-latency communication. We introduce $\textit{Bifröst}$, a programming model that realizes spatial networking using bigraphs to express both containment and connectivity, enabling policies to be scoped by physical boundaries, devices to be named by location, the instantiation of spatial services, and the composition of spaces while maintaining local autonomy. Bifröst enables a new class of spatially-aware applications, where co-located devices communicate directly, physical barriers require explicit gateways, and local control bridges to global coordination.

The desirability of the mid-band frequency range (1 - 10 GHz) for federal and commercial applications, combined with the growing applications for commercial indoor use-cases, such as factory automation, opens up a new approach to spectrum sharing: the same frequency bands used outdoors by federal incumbents can be reused by commercial indoor users. A recent example of such sharing, between commercial systems, is the 6 GHz band (5.925 - 7.125 GHz) where unlicensed, low-power-indoor (LPI) users share the band with outdoor incumbents, primarily fixed microwave links. However, to date, there exist no reliable, automatic means of determining whether a device is indoors or outdoors, necessitating the use of other mechanisms such as mandating indoor access points (APs) to have integrated antennas and not be battery powered, and reducing transmit power of client devices which may be outdoors. An accurate indoor/outdoor (I/O) classification addresses these challenges, enabling automatic transmit power adjustments without interfering with incumbents. To this end, we leverage the Global Navigation Satellite System (GNSS) signals for I/O classification. GNSS signals, designed inherently for outdoor reception and highly susceptible to indoor attenuation and blocking, provide a robust and distinguishing feature for environmental sensing. We develop various methodologies, including threshold-based techniques and machine learning approaches and evaluate them using an expanded dataset gathered from diverse geographical locations. Our results demonstrate that GNSS-based methods alone can achieve greater accuracy than approaches relying solely on wireless (Wi-Fi) data, particularly in unfamiliar locations. Furthermore, the integration of GNSS data with Wi-Fi information leads to improved classification accuracy, showcasing the significant benefits of multi-modal data fusion.

Estimating set similarity measures is a fundamental problem in data analysis, with applications in information retrieval, database systems, and streaming algorithms. Among such measures, the Jaccard containment of two sets A, B--defined as ϕ = | A B | |A | [0, 1] when A is treated as the reference set--is particularly important in asymmetric comparison tasks, such as detecting near-duplicates or containment-based joins. In large-scale settings, exact computation of ϕ may be infeasible, as it requires full knowledge of both sets. Sampling-based estimators that use small random subsets P A and Q B can be used as scalable alternatives when the sizes |A |, |B | are known, such as in Oracle databases. This paper presents a theoretical analysis of the likelihood models and estimation strategies for Jaccard containment based on random samples, focusing on both empirical performance and statistical guarantees.

Misinformation is still a major societal problem and the arrival of Large Language Models (LLMs) only added to it. This paper analyzes synthetic, false, and genuine information in the form of text from spectral analysis, visualization, and explainability perspectives to find the answer to why the problem is still unsolved despite multiple years of research and a plethora of solutions in the literature. Various embedding techniques on multiple datasets are used to represent information for the purpose. The diverse spectral and non-spectral methods used on these embeddings include t-distributed Stochastic Neighbor Embedding (t-SNE), Principal Component Analysis (PCA), and Variational Autoencoders (VAEs). Classification is done using multiple machine learning algorithms. Local Interpretable Model-Agnostic Explanations (LIME), SHapley Additive exPlanations (SHAP), and Integrated Gradients are used for the explanation of the classification. The analysis and the explanations generated show that misinformation is quite closely intertwined with genuine information and the machine learning algorithms are not as effective in separating the two despite the claims in the literature.

Recent studies on the Shapes Constraint Language (SHACL), a W3C specification for validating RDF graphs, rely on translating the language into first-order logic in order to provide formally-grounded solutions to the validation, containment and satisfiability decision problems. Continuing on this line of research, we introduce SHACL2FOL, the first automatic tool that (i) translates SHACL documents into FOL sentences and (ii) computes the answer to the two static analysis problems of satisfiability and containment; it also allow to test the validity of a graph with respect to a set of constraints. By integrating with existing theorem provers, such as E and Vampire, the tool computes the answer to the aforementioned decision problems and outputs the corresponding first-order logic theories in the standard TPTP format. We believe this tool can contribute to further theoretical studies of SHACL, by providing an automatic first-order logic interpretation of its semantics, while also benefiting SHACL practitioners, by supplying static analysis capabilities to help the creation and management of SHACL constraints.

We present an in-depth analysis of data discovery in data lakes, focusing on table augmentation for given machine learning tasks. We analyze alternative methods used in the three main steps: retrieving joinable tables, merging information, and predicting with the resultant table. As data lakes, the paper uses YADL (Yet Another Data Lake) -- a novel dataset we developed as a tool for benchmarking this data discovery task -- and Open Data US, a well-referenced real data lake. Through systematic exploration on both lakes, our study outlines the importance of accurately retrieving join candidates and the efficiency of simple merging methods. We report new insights on the benefits of existing solutions and on their limitations, aiming at guiding future research in this space.

In a quaint Regency-era office overlooking London's Russell Square, I cofounded a company called DeepMind with two friends, Demis Hassabis and Shane Legg, in the summer of 2010. Our goal, one that still feels as ambitious and crazy and hopeful as it did back then, was to replicate the very thing that makes us unique as a species: our intelligence. To achieve this, we would need to create a system that could imitate and then eventually outperform all human cognitive abilities, from vision and speech to planning and imagination, and ultimately empathy and creativity. Since such a system would benefit from the massively parallel processing of supercomputers and the explosion of vast new sources of data from across the open web, we knew that even modest progress toward this goal would have profound societal implications. It certainly felt pretty far-out at the time.

What is it with wave metaphors? Technological determinists – people who believe that technology drives history – love them. Think of Alvin Toffler, who saw the history of civilisation as a succession of three such waves (agricultural, industrial and post-industrial). The idea is of immense power, unstoppable, moving inexorably towards us as we cower before its immensity, much as the dinosaurs must have done when they saw the mile-high tsunami heading in their direction. Mustafa Suleyman says he is not a determinist, but at times he sounds awfully like one.

[HHM20] discovered, for 7 pairs (C,D) of seemingly distinct standard electoral control types, that C and D are identical: For each input I and each election system, I is a Yes instance of both C and D, or of neither. Surprisingly this had gone undetected, even as the field was score-carding how many std. control types election systems were resistant to; various "different" cells on such score cards were, unknowingly, duplicate effort on the same issue. This naturally raises the worry that other pairs of control types are also identical, and so work still is being needlessly duplicated. We determine, for all std. control types, which pairs are, for elections whose votes are linear orderings of the candidates, always identical. We show that no identical control pairs exist beyond the known 7. We for 3 central election systems determine which control pairs are identical ("collapse") with respect to those systems, and we explore containment/incomparability relationships between control pairs. For approval voting, which has a different "type" for its votes, [HHM20]'s 7 collapses still hold. But we find 14 additional collapses that hold for approval voting but not for some election systems whose votes are linear orderings. We find 1 additional collapse for veto and none for plurality. We prove that each of the 3 election systems mentioned have no collapses other than those inherited from [HHM20] or added here. But we show many new containment relationships that hold between some separating control pairs, and for each separating pair of std. control types classify its separation in terms of containment (always, and strict on some inputs) or incomparability. Our work, for the general case and these 3 important election systems, clarifies the landscape of the 44 std. control types, for each pair collapsing or separating them, and also providing finer-grained information on the separations.