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.
Claims about AI safety reach audiences well beyond the AI community, yet many rely on opaque evidence or static assessments, when supporting evidence is accessible at all. We present the Systemic Risk...
The article discusses how automated red‑teaming can uncover more vulnerabilities at lower cost than human red‑teaming on AI safety benchmarks, yet this comparison conflates measurement with conclusion. It argues that benchmarks only assess harms within a predefined set, leaving a "threat‑model coverage gap" that can hide new risks, as seen in non‑English prompts. The authors suggest that evaluators from deployment contexts distinct from developers are needed to close this gap.
arXiv:2607. 07474v1 Announce Type: cross Abstract: Agentic red-teaming benchmarks report whether an injected agent was compromised as a single bit: the attack succeeded, or it did not.
arXiv:2608. 07446v1 Announce Type: cross Abstract: Rapid adoption of large language models (LLMs) in enterprise settings has introduced operational, security, and governance risks.
The paper introduces Governance-as-Code (GaC), a framework that translates the EU AI Act’s technical requirements into 43 machine‑checkable acceptance criteria across six compliance modules. GaC runs within a CI/CD pipeline, producing Article‑indexed audit evidence and providing actual Rego policy code. The authors validate GaC on two enterprise deployments, showing it reproduces manual audit findings—including three penalty‑triggering violations—while reducing audit labor by about 75%.
arXiv:2605. 28591v2 Announce Type: replace-cross Abstract: The validity of AI safety evaluations depends on models behaving consistently across controlled and deployment settings.
The paper introduces the AI Assessment Sandbox Configurator, an open‑source framework designed to support technical assessment in AI Regulatory Sandboxes (AIRS) mandated by the EU Artificial Intelligence Act. It outlines 11 architectural and governance requirements for infrastructure that enables large‑scale, structured technical testing, and presents a catalogue of tests, a shared data model, dashboards, and reporting tools that harmonise heterogeneous outputs. An early‑stage pilot demonstrated the framework’s harmonisation and reporting capabilities within a live AIRS engagement, contributing to an official Exit Report.
arXiv:2606. 30219v1 Announce Type: new Abstract: LLM evaluation and AI safety face a shared measurement problem: benchmark scores, reward-model signals, and reported safety metrics can improve while the latent properties they are meant to represent remain difficult to verify.
arXiv:2609.13642v1 Announce Type: new Abstract: We argue that a recurring failure in the evaluation of deployed AI systems occurs when data collected for operational monitoring or regulatory complian...
arXiv:2607. 16112v1 Announce Type: new Abstract: Frontier AI companies have published capability thresholds that differ substantially, making it difficult for third parties to verify whether a threshold has been crossed or to compare requirements across companies.
arXiv:2606. 09809v1 Announce Type: new Abstract: AI evaluation results are produced at scale but reported inconsistently across leaderboards, model cards, benchmark papers, and company blogs.
The paper investigates whether existing AI safety benchmarks, designed for large language models, are suitable for evaluating small language models (SLMs). By testing five benchmark suites on 26 open‑source SLMs with a unified scoring rubric, the authors find that ambiguous judgments dominate, especially for complex prompts and certain architectures. This ambiguity, linked to factors like lexical density and output perplexity, undermines the reliability of aggregate leaderboards and reveals a confound between model capability and perceived safety.