AI and AI Policies

Auditing for Bias in Ad Delivery Using Inferred Demographic Attributes Basileal Imana, Aleksandra Korolova, John Heidemann Auditing social-media algorithms has become a focus of public-interest research and policymaking to ensure their fairness across demographic groups such as race, age, and gender in consequential domains such as the presentation of employment opportunities. However, such demographic attributes are often unavailable to auditors and platforms. When demographics data is unavailable, auditors commonly infer them from other available information. In this work, we study the effects of inference error on auditing for bias in one prominent application: black-box audit of ad delivery using paired ads. We show that inference error, if not accounted for, causes auditing to falsely miss skew that exists. We then propose a way to mitigate the inference error when evaluating skew in ad delivery algorithms. Our method works by adjusting for expected error due to demographic inference, and it makes skew detection more sensitive when attributes must be inferred. Because inference is increasingly used for auditing, our results provide an important addition to the auditing toolbox to promote correct audits of ad delivery algorithms for bias. While the impact of attribute inference on accuracy has been studied in other domains, our work is the first to consider it for black-box evaluation of ad delivery bias, when only aggregate data is available to the auditor. Published in ACM Conference on Fairness, Accountability, and Transparency 2025 (ACM FAccT 2025) What the AI Whistleblower Protection Act Would Mean for Tech Workers" Written by Sophie Luskin Featured in Tech Policy Press Dynamic Risk Assessments for Offensive Cybersecurity Agents Comments of Researchers at Princeton University on the 2025 National AI R&D Strategic Plan, NSF-2025-OGC-0001 … Our comment makes three core points. First, the government should prioritize promoting the diffusion of AI-related technologies. Specifically, this requires prioritizing research to aid public sector and government applications of AI. In addition, research funding should support the infrastructure, capabilities, and institutional adaptations that enable productive AI adoption. Second, the government should invest in supporting the development of open models to democratize access to technology. This includes research infrastructure for the AI communities at universities. Third, the government should prioritize research on the impact of AI on the workforce, anticipate potential disruptions, and develop strategies to address them….