arXiv subjects

Robert Mahari

Publications and source records attributed to Robert Mahari.

At least 19 recordsLinked to original sources

arXiv2026

AISPA: User-Centric System Prompt Auditing for Large Language Model Applications

System prompts are instructions configured by developers to govern the behaviors of foundation models in AI applications. They are used throughout commercial AI products, but are rarely disclosed to the public or regulators, creating a serious trust and accountability gap in the wide deployment of AI systems. In this paper, we introduce Artificial Intelligence System Prompt Assurance (AISPA), a user-centric framework for systematically auditing system prompts in AI systems. AISPA examines specific parts of a system prompt and evaluates them along eight dimensions that matter to users. We then use this framework to review 3,249 instructions from system prompts in 88 commercial AI products, classifying each instruction as either protective (of users) or problematic. Our audit surfaces four core findings. First, system prompt design varies substantially across products and developers, with some organizations averaging over 60 protective instructions per product while others average fewer than 5. Second, protective instructions are widely adopted but shallow in scope: 98.9% of products contain at least one, yet only 24% cover all eight dimensions of the AISPA taxonomy. Third, system prompts have grown steadily longer and more protective of users, suggesting that user protection is becoming a more visible concern in commercial prompt design. Fourth, despite this progress, problematic instructions remain pervasive: roughly 40% of products contain at least one instruction that works against user interests, and protective and problematic instructions frequently coexist within the same prompt. Our findings highlight the need for greater transparency, standardization, and independent oversight for system prompts in commercial AI products.

cs.AI↗

arXiv2026

Limits of Predictability in Civil Litigation

Legal practice routinely relies on informal assessments of case strength, yet no large-scale empirical benchmark exists for how predictable civil-litigation outcomes actually are. Civil litigation unfolds through sequential filings, and parties may settle at any stage, yet most computational studies of legal prediction observe cases only after resolution, leaving open whether outcomes are predictable beforehand. Using 102{,}721 U.S.\ civil cases and 835{,}190 court filings from 1996 to 2022, we model each case as it evolves, predicting plaintiff win, plaintiff loss, or settlement at each stage from structured, textual, and institutional features available up to that point. The classifier achieves class-specific AUC values of 0.74--0.81 and up to 97\% accuracy for high-confidence predictions, providing a large-scale benchmark for litigation predictability before resolution. We characterize heterogeneity in predictability using case complexity, defined as the entropy of the predicted outcome distribution. Complexity is systematically higher in cases involving corporate parties and in cases only weakly anchored to precedent. Richer information improves prediction mainly in low-complexity cases, with diminishing returns as complexity rises: some disputes are hard to predict not for lack of information, but because their outcomes are genuinely less determinate. Complexity also rises as litigation progresses, indicating that additional filings can sustain or amplify uncertainty rather than resolve it. Settlement rates follow an inverted U-shape in complexity, peaking at intermediate uncertainty and declining at both extremes. These findings suggest that predictive uncertainty is not mere model error, but a structured signal of legal complexity, litigation dynamics, and how disputes are resolved.

cs.SI↗

arXiv2026

Intentionality is a Design Decision: Measuring Functional Intentionality for Accountable AI Systems

As AI systems increasingly exhibit autonomous, goal-directed, and long-horizon behavior, users lack a standardized way to detect the degree to which a system functions like an intentional actor for governance and accountability purposes. This position paper defines intentionality not as consciousness, but as a behavioral profile characterized by purpose, foresight, volition, temporal commitment, and coherence - criteria long used in legal and philosophical contexts to infer intent. These properties are design-contingent: architectural choices such as memory persistence, planning depth, and tool autonomy shape the degree to which systems exhibit organized goal pursuit. If intentionality is design-contingent, it is in principle controllable. Yet control requires measurement. We introduce the Functional Intentionality Test (FIT), a multidimensional framework that quantifies intentional-like behavior across five observable dimensions, and propose FIT-Eval, a structured evaluation protocol for eliciting and scoring them. While reduced human agency can increase efficiency, rising intentional capacity heightens accountability risks. By translating intentionality into interpretable levels, FIT enables proportionate oversight and deliberate autonomy calibration in increasingly agentic systems.

cs.AI↗

arXiv2026

The Determinants of Judicial Promotion: Politics, Prestige, and Performance

Judicial promotions shape the composition of higher courts, yet their determinants remain poorly understood. This paper examines promotion from U.S. District Courts to Courts of Appeals using a discrete-time hazard framework that models annual promotion probability. Using a judge-year panel covering over 36,000 observations from 1930 to present, we incorporate career timing, political alignment, elite credentials, and judicial performance measures. Promotion probabilities follow a life-cycle pattern and are strongly influenced by political alignment between judges and presidents ($\beta$ = 2.12, p < 0.001). Elite credentials and productivity increase promotion likelihood, while higher reversal rates reduce it. Citation network centrality exhibits a meaningful association ($\beta$ = 0.230, p = 0.025) that operates independently of elite credentials. Promotion outcomes reflect a dynamic process shaped by timing, politics, elite networks, and performance signals, with political considerations dominating but not eclipsing judicial behavior.

cs.SI↗

arXiv2025

Community-centric modeling of citation dynamics explains collective citation patterns in science, law, and patents

Many human knowledge systems, such as science, law, and invention, are built on documents and the citations that link them. Citations, while serving multiple purposes, primarily function as a way to explicitly document the use of prior work and thus have become central to the study of knowledge systems. Analyzing citation dynamics has revealed statistical patterns that shed light on knowledge production, recognition, and formalization, and has helped identify key mechanisms driving these patterns. However, most quantitative findings are confined to scientific citations, raising the question of universality of these findings. Moreover, existing models of individual citation trajectories fail to explain phenomena such as delayed recognition, calling for a unifying framework. Here, we analyze a newly available corpus of U.S. case law, in addition to scientific and patent citation networks, to show that they share remarkably similar citation patterns, including a heavy-tailed distribution of sleeping beauties. We propose a holistic model that captures the three core mechanisms driving collective dynamics and replicates the elusive phenomenon of delayed recognition. We demonstrate that the model not only replicates observed citation patterns, but also better predicts future successes by considering the whole system. Our work offers insights into key mechanisms that govern large-scale patterns of collective human knowledge systems and may provide generalizable perspectives on discovery and innovation across domains.

physics.soc-ph↗

arXiv2025

Authenticated Delegation and Authorized AI Agents

The rapid deployment of autonomous AI agents creates urgent challenges around authorization, accountability, and access control in digital spaces. New standards are needed to know whom AI agents act on behalf of and guide their use appropriately, protecting online spaces while unlocking the value of task delegation to autonomous agents. We introduce a novel framework for authenticated, authorized, and auditable delegation of authority to AI agents, where human users can securely delegate and restrict the permissions and scope of agents while maintaining clear chains of accountability. This framework builds on existing identification and access management protocols, extending OAuth 2.0 and OpenID Connect with agent-specific credentials and metadata, maintaining compatibility with established authentication and web infrastructure. Further, we propose a framework for translating flexible, natural language permissions into auditable access control configurations, enabling robust scoping of AI agent capabilities across diverse interaction modalities. Taken together, this practical approach facilitates immediate deployment of AI agents while addressing key security and accountability concerns, working toward ensuring agentic AI systems perform only appropriate actions and providing a tool for digital service providers to enable AI agent interactions without risking harm from scalable interaction.

cs.CY↗

arXiv2024

Bridging the Data Provenance Gap Across Text, Speech and Video

Progress in AI is driven largely by the scale and quality of training data. Despite this, there is a deficit of empirical analysis examining the attributes of well-established datasets beyond text. In this work we conduct the largest and first-of-its-kind longitudinal audit across modalities--popular text, speech, and video datasets--from their detailed sourcing trends and use restrictions to their geographical and linguistic representation. Our manual analysis covers nearly 4000 public datasets between 1990-2024, spanning 608 languages, 798 sources, 659 organizations, and 67 countries. We find that multimodal machine learning applications have overwhelmingly turned to web-crawled, synthetic, and social media platforms, such as YouTube, for their training sets, eclipsing all other sources since 2019. Secondly, tracing the chain of dataset derivations we find that while less than 33% of datasets are restrictively licensed, over 80% of the source content in widely-used text, speech, and video datasets, carry non-commercial restrictions. Finally, counter to the rising number of languages and geographies represented in public AI training datasets, our audit demonstrates measures of relative geographical and multilingual representation have failed to significantly improve their coverage since 2013. We believe the breadth of our audit enables us to empirically examine trends in data sourcing, restrictions, and Western-centricity at an ecosystem-level, and that visibility into these questions are essential to progress in responsible AI. As a contribution to ongoing improvements in dataset transparency and responsible use, we release our entire multimodal audit, allowing practitioners to trace data provenance across text, speech, and video.

cs.AI↗

arXiv2024

Early Career Citations Capture Judicial Idiosyncrasies and Predict Judgments

Judicial impartiality is a cornerstone of well-functioning legal systems. We assemble a dataset of 112,312 civil lawsuits in U.S. District Courts to study the effect of extraneous factors on judicial decision making. We show that cases are randomly assigned to judges and that biographical judge features are predictive of judicial decisions. We use low-dimensional representations of judges' early-career citation records as generic representations of judicial idiosyncrasies. These predict future judgments with accuracies exceeding 65% for high-confidence predictions on balanced out-of-sample test cases. For 6-8% of judges, these representations are significant predictors across all judgments. These findings indicate that a small but significant group of judges routinely relies on extraneous factors and careful vetting of judges prior to appointment may partially address this issue. Our use of low-dimensional representations of citation records may also be generalized to other jurisdictions or to study other aspects of judicial decision making.

cs.SI↗

arXiv2024

Data-Driven Law Firm Rankings to Reduce Information Asymmetry in Legal Disputes

Selecting capable counsel can shape the outcome of litigation, yet evaluating law firm performance remains challenging. Widely used rankings prioritize prestige, size, and revenue rather than empirical litigation outcomes, offering little practical guidance. To address this gap, we build on the Bradley-Terry model and introduce a new ranking framework that treats each lawsuit as a competitive game between plaintiff and defendant law firms. Leveraging a newly constructed dataset of 60,540 U.S. civil lawsuits involving 54,541 law firms, our findings show that existing reputation-based rankings correlate poorly with actual litigation success, whereas our outcome-based ranking substantially improves predictive accuracy. These findings establish a foundation for more transparent, data-driven assessments of legal performance.

cs.CY↗

arXiv2024

Consent in Crisis: The Rapid Decline of the AI Data Commons

General-purpose artificial intelligence (AI) systems are built on massive swathes of public web data, assembled into corpora such as C4, RefinedWeb, and Dolma. To our knowledge, we conduct the first, large-scale, longitudinal audit of the consent protocols for the web domains underlying AI training corpora. Our audit of 14,000 web domains provides an expansive view of crawlable web data and how codified data use preferences are changing over time. We observe a proliferation of AI-specific clauses to limit use, acute differences in restrictions on AI developers, as well as general inconsistencies between websites' expressed intentions in their Terms of Service and their robots.txt. We diagnose these as symptoms of ineffective web protocols, not designed to cope with the widespread re-purposing of the internet for AI. Our longitudinal analyses show that in a single year (2023-2024) there has been a rapid crescendo of data restrictions from web sources, rendering ~5%+ of all tokens in C4, or 28%+ of the most actively maintained, critical sources in C4, fully restricted from use. For Terms of Service crawling restrictions, a full 45% of C4 is now restricted. If respected or enforced, these restrictions are rapidly biasing the diversity, freshness, and scaling laws for general-purpose AI systems. We hope to illustrate the emerging crises in data consent, for both developers and creators. The foreclosure of much of the open web will impact not only commercial AI, but also non-commercial AI and academic research.

cs.CL↗

arXiv2024

Data Authenticity, Consent, & Provenance for AI are all broken: what will it take to fix them?

New capabilities in foundation models are owed in large part to massive, widely-sourced, and under-documented training data collections. Existing practices in data collection have led to challenges in tracing authenticity, verifying consent, preserving privacy, addressing representation and bias, respecting copyright, and overall developing ethical and trustworthy foundation models. In response, regulation is emphasizing the need for training data transparency to understand foundation models' limitations. Based on a large-scale analysis of the foundation model training data landscape and existing solutions, we identify the missing infrastructure to facilitate responsible foundation model development practices. We examine the current shortcomings of common tools for tracing data authenticity, consent, and documentation, and outline how policymakers, developers, and data creators can facilitate responsible foundation model development by adopting universal data provenance standards.

cs.AI↗

arXiv2024

Insights from an experiment crowdsourcing data from thousands of US Amazon users: The importance of transparency, money, and data use

Data generated by users on digital platforms are a crucial resource for advocates and researchers interested in uncovering digital inequities, auditing algorithms, and understanding human behavior. Yet data access is often restricted. How can researchers both effectively and ethically collect user data? This paper shares an innovative approach to crowdsourcing user data to collect otherwise inaccessible Amazon purchase histories, spanning 5 years, from more than 5000 US users. We developed a data collection tool that prioritizes participant consent and includes an experimental study design. The design allows us to study multiple aspects of privacy perception and data sharing behavior. Experiment results (N=6325) reveal both monetary incentives and transparency can significantly increase data sharing. Age, race, education, and gender also played a role, where female and less-educated participants were more likely to share. Our study design enables a unique empirical evaluation of the "privacy paradox", where users claim to value their privacy more than they do in practice. We set up both real and hypothetical data sharing scenarios and find measurable similarities and differences in share rates across these contexts. For example, increasing monetary incentives had a 6 times higher impact on share rates in real scenarios. In addition, we study participants' opinions on how data should be used by various third parties, again finding demographics have a significant impact. Notably, the majority of participants disapproved of government agencies using purchase data yet the majority approved of use by researchers. Overall, our findings highlight the critical role that transparency, incentive design, and user demographics play in ethical data collection practices, and provide guidance for future researchers seeking to crowdsource user generated data.

cs.CY↗

arXiv2024

Leveraging Large Language Models for Learning Complex Legal Concepts through Storytelling

Making legal knowledge accessible to non-experts is crucial for enhancing general legal literacy and encouraging civic participation in democracy. However, legal documents are often challenging to understand for people without legal backgrounds. In this paper, we present a novel application of large language models (LLMs) in legal education to help non-experts learn intricate legal concepts through storytelling, an effective pedagogical tool in conveying complex and abstract concepts. We also introduce a new dataset LegalStories, which consists of 294 complex legal doctrines, each accompanied by a story and a set of multiple-choice questions generated by LLMs. To construct the dataset, we experiment with various LLMs to generate legal stories explaining these concepts. Furthermore, we use an expert-in-the-loop approach to iteratively design multiple-choice questions. Then, we evaluate the effectiveness of storytelling with LLMs through randomized controlled trials (RCTs) with legal novices on 10 samples from the dataset. We find that LLM-generated stories enhance comprehension of legal concepts and interest in law among non-native speakers compared to only definitions. Moreover, stories consistently help participants relate legal concepts to their lives. Finally, we find that learning with stories shows a higher retention rate for non-native speakers in the follow-up assessment. Our work has strong implications for using LLMs in promoting teaching and learning in the legal field and beyond.

cs.CL↗

arXiv2024

Verifiable evaluations of machine learning models using zkSNARKs

In a world of increasing closed-source commercial machine learning models, model evaluations from developers must be taken at face value. These benchmark results-whether over task accuracy, bias evaluations, or safety checks-are traditionally impossible to verify by a model end-user without the costly or impossible process of re-performing the benchmark on black-box model outputs. This work presents a method of verifiable model evaluation using model inference through zkSNARKs. The resulting zero-knowledge computational proofs of model outputs over datasets can be packaged into verifiable evaluation attestations showing that models with fixed private weights achieve stated performance or fairness metrics over public inputs. We present a flexible proving system that enables verifiable attestations to be performed on any standard neural network model with varying compute requirements. For the first time, we demonstrate this across a sample of real-world models and highlight key challenges and design solutions. This presents a new transparency paradigm in the verifiable evaluation of private models.

cs.LG↗

arXiv2023

zkTax: A pragmatic way to support zero-knowledge tax disclosures

Tax returns contain key financial information of interest to third parties: public officials are asked to share financial data for transparency, companies seek to assess the financial status of business partners, and individuals need to prove their income to landlords or to receive benefits. Tax returns also contain sensitive data such that sharing them in their entirety undermines privacy. We introduce a zero-knowledge tax disclosure system (zkTax) that allows individuals and organizations to make provable claims about select information in their tax returns without revealing additional information, which can be independently verified by third parties. The system consists of three distinct services that can be distributed: a tax authority provides tax documents signed with a public key; a Redact & Prove Service enables users to produce a redacted version of the tax documents with a zero-knowledge proof attesting the provenance of the redacted data; a Verify Service enables anyone to verify the proof. We implement a prototype with a user interface, compatible with U.S. tax forms, and demonstrate how this design could be implemented with minimal changes to existing tax infrastructure. Our system is designed to be extensible to other contexts and jurisdictions. This work provides a practical example of how distributed tools leveraging cryptography can enhance existing government or financial infrastructures, providing immediate transparency alongside privacy without system overhauls.

cs.CR↗

arXiv2023

LePaRD: A Large-Scale Dataset of Judges Citing Precedents

We present the Legal Passage Retrieval Dataset LePaRD. LePaRD is a massive collection of U.S. federal judicial citations to precedent in context. The dataset aims to facilitate work on legal passage prediction, a challenging practice-oriented legal retrieval and reasoning task. Legal passage prediction seeks to predict relevant passages from precedential court decisions given the context of a legal argument. We extensively evaluate various retrieval approaches on LePaRD, and find that classification appears to work best. However, we note that legal precedent prediction is a difficult task, and there remains significant room for improvement. We hope that by publishing LePaRD, we will encourage others to engage with a legal NLP task that promises to help expand access to justice by reducing the burden associated with legal research. A subset of the LePaRD dataset is freely available and the whole dataset will be released upon publication.

cs.CL↗

arXiv2023

The Data Provenance Initiative: A Large Scale Audit of Dataset Licensing & Attribution in AI

The race to train language models on vast, diverse, and inconsistently documented datasets has raised pressing concerns about the legal and ethical risks for practitioners. To remedy these practices threatening data transparency and understanding, we convene a multi-disciplinary effort between legal and machine learning experts to systematically audit and trace 1800+ text datasets. We develop tools and standards to trace the lineage of these datasets, from their source, creators, series of license conditions, properties, and subsequent use. Our landscape analysis highlights the sharp divides in composition and focus of commercially open vs closed datasets, with closed datasets monopolizing important categories: lower resource languages, more creative tasks, richer topic variety, newer and more synthetic training data. This points to a deepening divide in the types of data that are made available under different license conditions, and heightened implications for jurisdictional legal interpretations of copyright and fair use. We also observe frequent miscategorization of licenses on widely used dataset hosting sites, with license omission of 70%+ and error rates of 50%+. This points to a crisis in misattribution and informed use of the most popular datasets driving many recent breakthroughs. As a contribution to ongoing improvements in dataset transparency and responsible use, we release our entire audit, with an interactive UI, the Data Provenance Explorer, which allows practitioners to trace and filter on data provenance for the most popular open source finetuning data collections: www.dataprovenance.org.

cs.CL↗

arXiv2023

The Law and NLP: Bridging Disciplinary Disconnects

Legal practice is intrinsically rooted in the fabric of language, yet legal practitioners and scholars have been slow to adopt tools from natural language processing (NLP). At the same time, the legal system is experiencing an access to justice crisis, which could be partially alleviated with NLP. In this position paper, we argue that the slow uptake of NLP in legal practice is exacerbated by a disconnect between the needs of the legal community and the focus of NLP researchers. In a review of recent trends in the legal NLP literature, we find limited overlap between the legal NLP community and legal academia. Our interpretation is that some of the most popular legal NLP tasks fail to address the needs of legal practitioners. We discuss examples of legal NLP tasks that promise to bridge disciplinary disconnects and highlight interesting areas for legal NLP research that remain underexplored.

cs.CL↗