Overview

Algorithmic auditing studies how to test automated decisions and the systems around them, especially when errors or disparities affect people. Common approaches compare subgroup outcomes, use counterfactual or causal tests, probe model behavior under perturbations, and apply statistical methods such as sequential testing and uncertainty analysis. Audits increasingly examine deployed pipelines, data and component interactions, strategic responses, and evidence trails—not just a model’s aggregate accuracy. A central challenge is making conclusions reliable and meaningful despite limited access, shifting conditions, incomplete data, and metrics that can be gamed or fail to capture the harms that matter.

Latest papers

159 papers20 papers in the last four weeks, down 9% on the four weeks before. 0.2% of all new papers.

Feb 27, 2026

A multimodal slice discovery framework for systematic failure detection and explanation in medical image classification

Despite advances in machine learning-based medical image classifiers, the safety and reliability of these systems remain major concerns in practical settings. Existing auditing approaches mainly rely on unimodal features or metadata-based subgroup analyses, which are limited in interpretability and often fail to capture hidden systematic failures. To address these limitations, we introduce the first automated auditing framework that extends slice discovery methods to multimodal representations specifically for medical applications. Comprehensive experiments were conducted under common failure scenarios using the MIMIC-CXR-JPG dataset, demonstrating the framework's strong capability in both failure discovery and explanation generation. Our results also show that multimodal information generally allows more comprehensive and effective auditing of classifiers, while unimodal variants beyond image-only inputs exhibit strong potential in scenarios where resources are constrained.
Feb 25, 2026

Global Sequential Testing for Multi-Stream Auditing

Across many risk-sensitive areas, it is critical to continuously audit machine learning systems as we receive more data to quickly determine if they are performing as designed. This auditing task can be modeled as a sequential hypothesis testing problem with kk data streams and a global null hypothesis that asserts the system operates as intended across all kk streams. Under the alternative, the standard global sequential test, which uses a Bonferroni correction, has an expected stopping time of O(ln⁡kα)O\left(\ln \frac{k}α\right) for large kk and significance level αα. In this work, we demonstrate that efficient sequential tests, relying on merging martingales via averaging and products rules, provide improved stopping times, and thus more powerful tests against the null. Using these results, we show that a balanced test can match the Bonferroni rate of O(ln⁡kα)O\left(\ln \frac{k}α\right) in the sparse regime (just a few non-null streams) while achieving O(1kln⁡1α)O\left(\frac{1}{k}\ln \frac{1}α\right) under dense alternatives (many non-null steams). We validate our theory through experiments on both synthetic and real-world data.
Feb 16, 2026

Governing AI Forgetting: Auditing for Machine Unlearning Compliance

Despite legal mandates for the right to be forgotten, AI operators routinely fail to comply with data deletion requests. While machine unlearning (MU) provides a technical solution to remove personal data's influence from trained models, ensuring compliance remains challenging due to the fundamental gap between MU's technical feasibility and regulatory implementation. In this paper, we introduce the first economic framework for auditing MU compliance, by integrating certified unlearning theory with regulatory enforcement. We first characterize MU's inherent verification uncertainty using a hypothesis-testing interpretation of certified unlearning to derive the auditor's detection capability, and then propose a game-theoretic model to capture the strategic interactions between the auditor and the operator. A key technical challenge arises from MU-specific nonlinearities inherent in the model utility and the detection probability, which create complex strategic couplings that traditional auditing frameworks do not address and that also preclude closed-form solutions. We address this by transforming the complex bivariate nonlinear fixed-point problem into a tractable univariate auxiliary problem, enabling us to decouple the system and establish the equilibrium existence, uniqueness, and structural properties without relying on explicit solutions. Counterintuitively, our analysis reveals that the auditor can optimally reduce the inspection intensity as deletion requests increase, since the operator's weakened unlearning makes non-compliance easier to detect. This is consistent with recent auditing reductions in China despite growing deletion requests. Moreover, we prove that although undisclosed auditing offers informational advantages for the auditor, it paradoxically reduces the regulatory cost-effectiveness relative to disclosed auditing.
Jan 23, 2026

Falsifying Discriminant Validity of Predictive Algorithms

Empirical investigations into unintended model behavior often show that the algorithm is predicting another outcome than what was intended. These exposés highlight the need to identify when algorithms predict unintended quantities - ideally before deploying them into consequential settings. We propose a falsification framework that provides a principled statistical test for discriminant validity: the requirement that an algorithm predict intended outcomes better than impermissible ones. Drawing on falsification practices from causal inference, econometrics, and psychometrics, our framework compares calibrated prediction losses across outcomes to assess whether the algorithm exhibits discriminant validity with respect to a specified impermissible proxy. In settings where the target outcome is difficult to observe, multiple permissible proxy outcomes may be available; our framework accommodates both this setting and the case with a single permissible proxy. Throughout we use nonparametric hypothesis testing methods that make minimal assumptions on the data-generating process. We illustrate the method in an admissions setting, where the framework establishes discriminant validity with respect to gender but fails to establish discriminant validity with respect to race. This demonstrates how falsification can serve as an early validity check. We also provide analysis in a criminal justice setting, where we highlight the limitations of our framework and emphasize the need for complementary approaches to assess other aspects of construct validity and external validity.
Jan 23, 2026

White-Box Sensitivity Auditing with Steering Vectors

Algorithmic audits are essential tools for examining systems for properties required by regulators or desired by operators. Current audits of large language models (LLMs) primarily rely on black-box evaluations that assess model behavior only through input-output testing. These methods are limited to tests constructed in the input space, often generated by heuristics. In addition, many socially relevant model properties (e.g., gender bias) are abstract and difficult to measure through text-based inputs alone. To address these limitations, we propose a white-box sensitivity auditing framework for LLMs that leverages activation steering to conduct more rigorous assessments through model internals. Our auditing method conducts internal sensitivity tests by manipulating key concepts relevant to the model's intended function for the task. We demonstrate its application to bias audits in four simulated high-stakes LLM decision tasks. Our method consistently indicates substantial dependence on protected attributes in model predictions, even in settings where standard black-box evaluations suggest little or no bias. Our code is openly available at https://github.com/hannahxchen/llm-steering-audit
Dec 15, 2025

One Permutation Is All You Need: Fast, Deterministic Feature Importance and Model Stress-Testing

Reliable estimation of feature contributions in machine learning models is essential for transparency, algorithmic fairness, and regulatory compliance. While permutation feature importance is widely used, classical implementations rely on repeated Monte Carlo shuffling, introducing significant computational overhead and stochastic instability. In this paper, we show that replacing BB random permutations with a single, max-min rank-optimal deterministic permutation maintains or improves correlation with ground-truth importance while eliminating estimation variance and reducing complexity from O(B⋅n⋅p)O(B \cdot n \cdot p) to O(n⋅p)O(n \cdot p). Under location-scale feature distributions, we formally prove exact recovery of scale-adjusted linear regression coefficients, alongside improved importance estimation under concave model sensitivity. We extend this deterministic framework along two complementary dimensions. First, Systemic Feature Importance (SFI) integrates empirical feature correlations to quantify indirect feature reliance through proxy variables. Second, Importance Direction extends scalar importance to a signed, directional representation by measuring concordance between covariate displacements and output shifts. Extensive empirical validation across nearly 200 simulation scenarios demonstrates superior bias-variance trade-offs in high-dimensional and low signal-to-noise regimes. Finally, two real-world credit risk case studies show how coupling SFI with Importance Direction enables practitioners and regulators to audit models for both the magnitude and net sign of hidden reliance on protected attributes, delivering a principled, transparent, and scalable framework for model governance.
Dec 5, 2025

A framework for auditing grounding claims

The symbol grounding problem asks how a token such as cat can be about cats. We propose a framework for auditing grounding claims against a declared semantic standard. The audit reports measurements and evidence, with overall verdicts conditional on explicit acceptance criteria. Its profiles assess accuracy, robustness, and composition alongside evidence about how the system acquired its mechanisms, how they contribute to performance, and why they were retained. In a toy gridworld, an agent interprets individual symbols accurately but fails a withheld combination. Composing its interpretations by the declared rule would succeed. This comparison identifies a departure from the composition rule within the observed failure. Both this audit and a pilot on pretrained word vectors provide evidence that a designated mechanism contributes to present performance. Whether that contribution explains its retention remains uncertified. The framework evaluates the evidence for grounding claims; candidate accounts remain responsible for explaining how meaning emerges.
Oct 5, 2025

A global log for medical AI

Modern computer systems rely on syslog, a universal protocol that records critical events across heterogeneous infrastructure. Medicine's rapidly growing AI stack has no equivalent. As medicine deploys AI tools at scale, there is no standard way to record how, when, by whom, and for whom these models are used. Without such records, it is difficult to measure real-world performance and outcomes, detect adverse events, or identify bias and dataset drift. Here we introduce MedLog, a protocol for event-level logging of medical AI. Each time an AI model interacts with a human, another algorithm, or an automated workflow, MedLog creates a record. Each record contains nine core fields: header, model, user, target, inputs, artifacts, outputs, outcomes, and feedback. We apply MedLog across four deployments in the US, Switzerland, and Vietnam: ICU deterioration prediction, tetanus progression monitoring from wearable signals, automated sepsis quality reporting, and patient attendance prediction. MedLog records capture model behavior, workflow interactions, and downstream outcomes, including AI performance degradation during severe weather events in patient attendance prediction and increased laboratory testing after ICU deterioration alerts. MedLog limits the data footprint through risk-based sampling, lifecycle-aware retention policies, and write-behind caching, enabling deployment in low-resource settings. It also supports detailed traces for complex, agentic, or multi-stage workflows, creating a foundation for continuous monitoring, auditing, and improvement of medical AI.
Jun 28, 2025

How Reliable are Fairness Audits with Unreliable Data?

Fairness audits are a key component of responsible machine-learning deployment. Yet, audit-recommendation reliability under incomplete protected-label access is still poorly understood. In this work, we focused on protected-label missingness in fairness mitigation audits. We introduced a seed-calibrated stress test to separate missingness effects from seed-to-seed movement already present under complete labels. Across ACS/Folktables tasks, missingness settings that retain some protected labels usually do not move selected mitigation methods beyond a complete-label seed-to-seed baseline. At 00% protected-label access, candidates collapse to an empirical-risk-minimization baseline and deterministic tie-breaking rather than revealing a broad missingness effect. We also found that threshold optimization can turn fairness gains on a single protected axis into intersectional harm above a seed baseline, and this threshold-optimizer finding persists under random-forest validation. Overall, our results highlight that protected-label missingness should be reported with seed-null calibration, candidate-set context, and intersectional consequences before it is treated as evidence of audit fragility.