The paper proposes a framework that connects structured hazard analysis, component-level testing, and probabilistic system modelling to assess system-level harms from AI in complex sociotechnical systems. It demonstrates the approach using the UK's Real Time Gross Settlement system, showing how adversarial inputs to LLM-based trading can shift AI behaviour, reduce system resilience, and increase the likelihood of cascading bank failures. The framework aims to provide a traceable pathway from model behaviour to systemic outcomes, enabling evidence-based governance of AI in critical infrastructure.
arXiv:2410. 22526v2 Announce Type: replace Abstract: To effectively address potential harms from Artificial Intelligence (AI) systems, it is essential to identify and mitigate system-level hazards.
arXiv:2607. 19292v1 Announce Type: cross Abstract: Current AI safety discourse still focuses disproportionately on visible failures, including obvious harms, dramatic misuse, and hypothetical catastrophic scenarios.
arXiv:2607. 18243v1 Announce Type: new Abstract: Agentic AI is crossing trust boundaries faster than current risk models can represent.
arXiv:2607. 02197v1 Announce Type: cross Abstract: The society and emerging risk-based regulatory frameworks for AI underscore the need for rigorous risk assessment to ensure safe and reliable AI systems.
The paper examines how a breach of a single AI vendor—used by banks for fraud screening, credit decisions, AML triage, customer analytics, and internal support—can spread through operational, informational, and financial links, ultimately causing losses that resemble a traditional banking crisis. It introduces a four‑layer heterogeneous network linking AI vendors, banks, interbank exposures, and customer accounts, and presents CFC‑Prop, a stochastic epidemic‑and‑clearing model that reproduces heavy‑tailed loss distributions and sensitivity to patch latency on a synthetic dataset of 60 vendors, 220 banks, and 1,400 interbank exposures. Additionally, the authors develop CFC‑GNN, an early‑warning graph‑based model that predicts high‑cascade‑risk vendors with AUROC 0.82 and AUPRC 0.60, and they release all code, data, and scripts for reproducibility.
arXiv:2607. 23365v1 Announce Type: cross Abstract: Artificial intelligence (AI) systems are increasingly deployed in high-stakes domains such as healthcare, autonomous driving, finance, and education.
arXiv:2606. 14594v1 Announce Type: cross Abstract: AI-assisted software development has moved from line-level autocomplete to agents that can plan changes, edit files, and submit pull requests with limited human supervision.
The paper proposes AI Deployment Accountability Engineering (ADAE), a new subdiscipline focused on establishing measurable, continuous, and actionable accountability for AI systems once they are deployed. ADAE treats accountability as a deployment-layer property, aiming to ensure systems remain within acceptable risk limits, identify failure contexts, attribute failures across technical and human components, and translate technical failures into downstream consequences. The authors outline a research agenda built around four pillars—structured discovery of context-dependent failure modes, privacy-preserving accountability measurement, system-level risk analysis for agentic AI, and translation of technical failures into operational and institutional risks—to support timely intervention in safety-critical socio-technical environments.
arXiv:2607. 14353v1 Announce Type: cross Abstract: As automated decision-making and data-driven technologies pervade society and are used to manage consequential outcomes, understanding the technology's capabilities, limitations, and attendant risks in context requires analysis of full sociotechnical systems.
The paper introduces the Systemic Risk Index, an open pipeline and dashboard that aggregates evidence from 19 public AI benchmarks into four systemic‑risk categories defined by the EU GPAI Code of Practice. It evaluates 18 models using harm‑preserving perturbations and simulated deployment contexts, offering users the ability to switch between average and worst‑case aggregation and to trace each risk rating back to its benchmark evidence. The study finds that worst‑case scores can be 14 to 37 points lower than average scores, and that LLM judges agree with human graders at a level comparable to human‑human agreement.
arXiv:2607. 29405v1 Announce Type: new Abstract: Agentic AI systems act through multi-step trajectories that combine planning, tool use, memory, interaction, and adaptation.
arXiv:2608. 11344v1 Announce Type: cross Abstract: Financial institutions are delegating consequential decisions to agentic AI systems that decompose goals, coordinate models and tools, and act with little oversight.