Black box large language models (LLMs) make information extraction (IE) easy to configure, but hard to trust. Unlike traditional information extraction pipelines, the information "extracted" is not guaranteed to be grounded in the document. To prevent this, this paper introduces the notion of a "safe passage": context generated by the LLM that is both grounded in the document and consistent with the extracted information. This is operationalized via a three-step pipeline, SafePassage, which consists of: (1) an LLM extractor that generates structured entities and their contexts from a document, (2) a string-based global aligner, and (3) a scoring model. Results show that using these three parts in conjunction reduces hallucinations by up to 85% on information extraction tasks with minimal risk of flagging non-hallucinations. High agreement between the SafePassage pipeline and human judgments of extraction quality mean that the pipeline can be dually used to evaluate LLMs. Surprisingly, results also show that using a transformer encoder fine-tuned on a small number of task-specific examples can outperform an LLM scoring model at flagging unsafe passages. These annotations can be collected in as little as 1-2 hours.

We introduce DRBench, a benchmark for evaluating AI agents on complex, open-ended deep research tasks in enterprise settings. Unlike prior benchmarks that focus on simple questions or web-only queries, DRBench evaluates agents on multi-step queries (for example, ``What changes should we make to our product roadmap to ensure compliance with this standard?") that require identifying supporting facts from both the public web and private company knowledge base. Each task is grounded in realistic user personas and enterprise context, spanning a heterogeneous search space that includes productivity software, cloud file systems, emails, chat conversations, and the open web. Tasks are generated through a carefully designed synthesis pipeline with human-in-the-loop verification, and agents are evaluated on their ability to recall relevant insights, maintain factual accuracy, and produce coherent, well-structured reports. We release 15 deep research tasks across 10 domains, such as Sales, Cybersecurity, and Compliance. We demonstrate the effectiveness of DRBench by evaluating diverse DR agents across open- and closed-source models (such as GPT, Llama, and Qwen) and DR strategies, highlighting their strengths, weaknesses, and the critical path for advancing enterprise deep research. Code is available at https://github.com/ServiceNow/drbench.

The wide adoption of AI decision-making systems in critical domains such as criminal justice, loan approval, and hiring processes has heightened concerns about algorithmic fairness. As we often only have access to the output of algorithms without insights into their internal mechanisms, it was natural to examine how decisions would alter when auxiliary sensitive attributes (such as race) change. This led the research community to come up with counterfactual fairness measures, but how to evaluate the measure from available data remains a challenging task. In many practical applications, the target counterfactual measure is not identifiable, i.e., it cannot be uniquely determined from the combination of quantitative data and qualitative knowledge. This paper addresses this challenge using partial identification, which derives informative bounds over counterfactual fairness measures from observational data. We introduce a Bayesian approach to bound unknown counterfactual fairness measures with high confidence. We demonstrate our algorithm on the COMPAS dataset, examining fairness in recidivism risk scores with respect to race, age, and sex. Our results reveal a positive (spurious) effect on the COMPAS score when changing race to African-American (from all others) and a negative (direct causal) effect when transitioning from young to old age.

Large language models (LLMs) have been extensively used for legal judgment prediction tasks based on case reports and crime history. However, with a surge in the availability of large vision language models (VLMs), legal judgment prediction systems can now be made to leverage the images of the criminals in addition to the textual case reports/crime history. Applications built in this way could lead to inadvertent consequences and be used with malicious intent. In this work, we run an audit to investigate the efficiency of standalone VLMs in the bail decision prediction task. We observe that the performance is poor across multiple intersectional groups and models \textit{wrongly deny bail to deserving individuals with very high confidence}. We design different intervention algorithms by first including legal precedents through a RAG pipeline and then fine-tuning the VLMs using innovative schemes. We demonstrate that these interventions substantially improve the performance of bail prediction. Our work paves the way for the design of smarter interventions on VLMs in the future, before they can be deployed for real-world legal judgment prediction.

Artificial Intelligence has rapidly become a cornerstone technology, significantly influencing Europe's societal and economic landscapes. However, the proliferation of AI also raises critical ethical, legal, and regulatory challenges. The CERTAIN (Certification for Ethical and Regulatory Transparency in Artificial Intelligence) project addresses these issues by developing a comprehensive framework that integrates regulatory compliance, ethical standards, and transparency into AI systems. In this position paper, we outline the methodological steps for building the core components of this framework. Specifically, we present: (i) semantic Machine Learning Operations (MLOps) for structured AI lifecycle management, (ii) ontology-driven data lineage tracking to ensure traceability and accountability, and (iii) regulatory operations (RegOps) workflows to operationalize compliance requirements. By implementing and validating its solutions across diverse pilots, CERTAIN aims to advance regulatory compliance and to promote responsible AI innovation aligned with European standards.

Designing wise AI policy is a grand challenge for society. To design such policy, policymakers should place a premium on rigorous evidence and scientific consensus. While several mechanisms exist for evidence generation, and nascent mechanisms tackle evidence synthesis, we identify a complete void on consensus formation. In this position paper, we argue NeurIPS should actively catalyze scientific consensus on AI policy. Beyond identifying the current deficit in consensus formation mechanisms, we argue that NeurIPS is the best option due its strengths and the paucity of compelling alternatives. To make progress, we recommend initial pilots for NeurIPS by distilling lessons from the IPCC's leadership to build scientific consensus on climate policy. We dispel predictable counters that AI researchers disagree too much to achieve consensus and that policy engagement is not the business of NeurIPS. NeurIPS leads AI on many fronts, and it should champion scientific consensus to create higher quality AI policy.

The growth and permeation of artificial intelligence (AI) technologies across society has drawn focus to the ways in which the responsible use of these technologies can be facilitated through AI governance. Increasingly, large companies and governments alike have begun to articulate and, in some cases, enforce governance preferences through AI policy. Yet existing literature documents an unwieldy heterogeneity in ethical principles for AI governance, while our own prior research finds that discussions of the implications of AI policy are not yet present in the computer science (CS) curriculum. In this context, overlapping jurisdictions and even contradictory policy preferences across private companies, local, national, and multinational governments create a complex landscape for AI policy which, we argue, will require AI developers able adapt to an evolving regulatory environment. Preparing computing students for the new challenges of an AI-dominated technology industry is therefore a key priority for the CS curriculum. In this discussion paper, we seek to articulate a framework for integrating discussions on the nascent AI policy landscape into computer science courses. We begin by summarizing recent AI policy efforts in the United States and European Union. Subsequently, we propose guiding questions to frame class discussions around AI policy in technical and non-technical (e.g., ethics) CS courses. Throughout, we emphasize the connection between normative policy demands and still-open technical challenges relating to their implementation and enforcement through code and governance structures. This paper therefore represents a valuable contribution towards bridging research and discussions across the areas of AI policy and CS education, underlining the need to prepare AI engineers to interact with and adapt to societal policy preferences.

We study empirical scaling laws for language model merging measured by cross-entropy. Despite its wide practical use, merging lacks a quantitative rule that predicts returns as we add experts or scale the model size. We identify a compact power law that links model size and expert number: the size-dependent floor decreases with model capacity, while the merging tail exhibits clear diminishing returns in the number of experts. The law holds in-domain and cross-domain, tightly fits measured curves across diverse architectures and methods (Average, TA, TIES, DARE), and explains two robust regularities: most gains arrive early, and variability shrinks as more experts are included. Building on this, we present a simple theory that explains why gains fall roughly as 1/k and links the floor and tail to properties of the base model and the diversity across domains. This law enables predictive planning: estimate how many experts are needed to reach a target loss, decide when to stop adding experts, and trade off scaling the base model versus adding experts under a fixed budget--turning merging from heuristic practice into a computationally efficient, planable alternative to multitask training. This suggests a scaling principle for distributed generative AI: predictable gains can be achieved by composing specialists, offering a complementary path toward AGI-level systems.

Safety is a paramount concern in clinical chatbot applications, where inaccurate or harmful responses can lead to serious consequences. Existing methods--such as guardrails and tool calling--often fall short in addressing the nuanced demands of the clinical domain. In this paper, we introduce TACOS (TAxonomy of COmprehensive Safety for Clinical Agents), a fine-grained, 21-class taxonomy that integrates safety filtering and tool selection into a single user intent classification step. TACOS is a taxonomy that can cover a wide spectrum of clinical and non-clinical queries, explicitly modeling varying safety thresholds and external tool dependencies. To validate our taxonomy, we curate a TACOS-annotated dataset and perform extensive experiments. Our results demonstrate the value of a new taxonomy specialized for clinical agent settings, and reveal useful insights about train data distribution and pretrained knowledge of base models.

Reliable evaluation of large language models is essential to ensure their applicability in practical scenarios. Traditional benchmark-based evaluation methods often rely on fixed reference answers, limiting their ability to capture important qualitative aspects of generated responses. To address these shortcomings, we propose an integrated evaluation framework called \textit{self-refining descriptive evaluation with expert-driven diagnostics}, SPEED, which utilizes specialized functional experts to perform comprehensive, descriptive analyses of model outputs. Unlike conventional approaches, SPEED actively incorporates expert feedback across multiple dimensions, including hallucination detection, toxicity assessment, and lexical-contextual appropriateness. Experimental results demonstrate that SPEED achieves robust and consistent evaluation performance across diverse domains and datasets. Additionally, by employing relatively compact expert models, SPEED demonstrates superior resource efficiency compared to larger-scale evaluators. These findings illustrate that SPEED significantly enhances fairness and interpretability in LLM evaluations, offering a promising alternative to existing evaluation methodologies.