Transparency

This post is part 2 in a series about automated content moderation. Read the first post here . When whistleblower Frances Haugen leaked a set of documents from Meta in 2020, among the revelations was a jarring statistic: The company’s algorithms designed to detect terrorist content incorrectly deleted nonviolent Arabic-language content 77 percent of the time, while failing to detect hate speech under the company’s own policies in many instances. Meta’s own transparency report released later that

Disclosure Assistant is a brand new startup targeting one of the more manual and time-consuming processes in legal practice: reviewing bank statements. Aimed at legal professionals who regularly review financial […] The post Start Up Corner: Disclosure Assistant – Review bank statements faster appeared first on Legal IT Insider .

Medical imaging models are often deployed without the demographic, acquisition, and quality metadata needed for subgroup auditing. Once those metadata disappear, clinically critical failure modes can be masked by strong aggregate performance, and many robust-learning methods lose the group structure they rely on. We present CAPRA, a calibrated proxy-axis framework for hidden subgroup analysis under missing metadata. CAPRA predicts image-derived semantic axes, calibrates axis posteriors on a smal

Fully autonomous weapon systems need to comply with International Humanitarian Law and underlying ethical principles. This requires the ability to recognize not only objects or persons to be targeted but also protected persons or objects. Such sophisticated object classification abilities, if achievable at all, would have to utilize machine learning techniques. These come with well-known limitations to predictability, reliability and explainability. This article argues such limitations could be

CSET’s Mina Narayanan shared her expert insight in an article published by TechCrunch. The article explores the lack of transparency surrounding how the U.S. government evaluates and approves the public release of advanced AI models, including OpenAI’s Sol and Anthropic’s Fable. The post How did the government decide OpenAI’s frontier model was safe to release? appeared first on Center for Security and Emerging Technology .

This article was originally published by IAPP linked here. Vietnam’s approach to artificial intelligence regulations crosses many topics and sectors, with a common theme emerging: human‑centered, state‑supervised and legally accountable. The country’s policy direction is clearly reflected in its first standalone Law on Artificial Intelligence, which took effect March 2026. At the same time, Vietnam [...] The post Vietnam clarifies AI authorship, training data and copyright liability: A comparati

Opinion: AI policy researcher Keller Scholl argues that a marketplace of AI auditors will always prioritize speed and cost over safety

Decentralized federated learning (DFL) removes the central server by letting nodes exchange model updates through peer-to-peer gossip, but existing gossip-based methods often lack provenance finality and resilience to Byzantine or lazy participants. Ledger-assisted federated learning (FL) improves auditability, yet blockchains, shards, or settlement committees can reintroduce global coordination costs that conflict with DFL locality. This paper proposes \emph{gspDAG-FL}, a secure DFL framework t

Colorado watered down its AI legislation but may still face litigation from the Department of Justice.

In the span of two weeks, the White House issued two of the most ambitious artificial intelligence directives in American history. On June 2, President Donald Trump signed an executive order mandating rapid AI adoption and hardened cyber defense across the government. Three days later, National Security Presidential Memorandum 11 directed every element of the national security enterprise to accelerate AI adoption, anchored by four pillars: adoption, adaptation, assurance, and accountability.The

Commission Opinion on the assessment of the Code of Practice on Transparency of AI-generated content Anonymous (not verified) Thu, 07/09/2026 - 09:03 Commission and AI Board consider this voluntary code as an effective mean to facilitate compliance with the AI Act transparency obligations. On july 8, the Commission concluded that the Code of Practice on Transparency of AI-generated content adequately covers the obligations provided for in Articles 50(2), (4) and (5) AI Act and facilitates their

Automated segmentation of cervical-spine MRI is increasingly used in clinical workflows, yet no fairness audit exists for this anatomy. We show that auditing these segmentation tasks is complicated by a common property of modern segmentation datasets: expert-annotated gold labels are expensive, so abundant machine-generated (silver) labels are added to limit annotation cost. This matters because the reference used to judge a model can itself be biased. In this study, we present the first fairnes

Die Informationsfreiheit in Deutschland befindet sich in einer Krise. Bereits die Ampelregierung hatte ihr Versprechen nicht eingelöst, die Informationsfreiheitsgesetze zu einem Transparenzgesetz weiterzuentwickeln. Dieser Trend gipfelte im Papier des Koalitionsausschusses vom 2.7.2026. Im Rahmen einer Paketlösung hat die Bundesregierung angekündigt, das IFG umfassend zu ändern. Die politischen Vorschläge des Koalitionsausschusses sind aus unserer Sicht weitreichend, sodass die Informationsfreih

Modern machine learning (ML) increasingly relies on complex models whose behavior is difficult to characterize beyond empirical performance metrics. Across a wide range of tasks, including prediction, generation, and decision-making, models with similar empirical performance can exhibit markedly different properties in terms of their transparency, interpretability, robustness, fairness, privacy, and certifiability. This survey highlights how optimization- and certification-oriented reasoning can

Learn how OpenAI approaches government and national security partnerships, with principles for responsible AI use, democratic accountability, and public safety.

Prior work has shown that chain-of-thought (CoT) reasoning is often unfaithful: a model's stated reasoning does not reliably reflect the process that produced its output. Detecting unfaithfulness, though, requires controlled experimental interventions, which cannot be applied to evaluation transcripts after the fact. We turn instead to a more tractable question that has received less attention: whether the stated reasoning is logically consistent with the answer it accompanies. Unlike faithfulne

As AI-generated media flood our information ecosystems, detecting synthetic content has become an urgent regulatory challenge – in fact, not one challenge but many, as synthetic media breeds problems across a range of digital contexts, including deepfakes and disinformation, scamming, and content moderation. The EU's new "Code of Practice on Transparency of AI-Generated Content" – the first concrete articulation of Article 50(2) AI Act, the EU's approach to AI-content detection – gives sensible

Ransomware poses an escalating cybersecurity threat as attackers continuously modify behavioral patterns to evade static defenses. Although existing machine learning-based detectors often achieve strong predictive performance, they generally assume fixed training data and do not support the selective removal of previously learned samples. This limitation conflicts with privacy regulations such as the GDPR and CCPA, which require the removal of sensitive user data upon request. To address this ch

This blog post is part 1 of a 2-part series. The second part sets out recommendations for companies and policymakers. Six years ago—one month into a global pandemic—we argued that the automated moderation processes many platforms were rapidly adopting should be highly transparent, easily appealable, and temporary. We warned that "protocols adopted in times of crisis often persist when the crisis is over." That warning proved prescient. The use of automation and artificial intelligence (AI) to id

Chatbots powered by generative AI (e.g., OpenAI's ChatGPT and Google's Gemini) are increasingly being appropriated for emotional support and companionship. These tools offer a suite of security and privacy (S&P) controls, including model training opt-outs and memory toggles, yet how the presence of these controls influences users' attitudes toward emotionally sensitive disclosure remains understudied. We conducted a mixed-methods vignette study with 354 U.S. participants to examine how S&P contr

Lars Klingbeil (SPD) und Friedrich Merz (CDU) im Gespräch. – Alle Rechte vorbehalten: IMAGO / Political-Moments Verbrämt als „Bürokratierückbau“ plant die Regierungskoalition das Plattmachen der Informationsfreiheit. Mehr als hundert Organisationen fordern heute in einem offenen Brief, den tiefen Einschnitt in Transparenzrechte und Pressefreiheit zu verhindern.

Who is legally responsible when Artificial Intelligence causes harm? The issue took centre stage on Tuesday – day two of the first ever UN summit on AI governance, where leading experts warned of ...

Foundation Models for Electronic Health Records (FEMRs) are pretrained on large-scale structured patient data, enabling them to convert longitudinal patient trajectories into generalizable representations for diverse clinical prediction tasks. Despite their effectiveness, FEMRs remain black-box models, raising concerns about bias, interpretability, and clinical trust. To address this, we propose the first token-level explainability approach for FEMRs. We train a Transformer-based surrogate model

Robbyant, the embodied artificial intelligence arm of Chinese fintech giant Ant Group, launched a new vision model that it claims can help robots overcome a long-standing challenge: accurately perceiving glass, mirrors and transparent objects. The unit of Hangzhou-based Ant Group on Tuesday unveiled its next-generation spatial perception model, LingBot-Depth 2.0, alongside a new foundational visual model called LingBot-Vision, as AI labs race to equip machines with the “brains” required to...

Vascular computed tomography datasets are commonly annotated only once per scan, yielding the pervasive yet under addressed problem of single mask annotation noise. Existing solutions either require costly multirater fusion or are coupled with network training, preventing explicit auditing of where and why labels fail. We introduce a decoupled framework for single-mask annotation noise detection that leverages cross-sectional patch self-consistency to produce interpretable and auditable noise ev

Fact checkers should maintain high standards of accountability because they hold unique positions in society by verifying content that can influence political practices and society as a whole. To maintain these professional standards, fact-checking network organizations such as the International Fact-Checking Network (IFCN) and the European Fact-Checking Standards Network (EFCSN) have established codes of standards, and fact-checking organizations should comply with them in a substantive way. Th

Major platforms constantly claim to fight disinformation and support the fact checking community, but their transparency reports and the empirical evidence from a survey of expert fact checkers across 21 EU countries show a different reality. This study finds that despite commitments made under EU regulations, expert fact checkers remain largely peripheral actors within the existing platform governance framework, with limited insight into how their work influences platform decisions. The post Ac

Social media platforms are increasingly using community-based verification systems, such as Community Notes, and AI systems to flag and contextualize potentially misleading content at scale. While these approaches promise speed and broad coverage, concerns about accuracy, bias, and transparency persist. Drawing on interviews with 29 fact checkers, we find that practitioners see community-based verification and AI Note Writers as complementary tools that can support, but not replace, professional

Deep learning has advanced cancer imaging and cancer-related risk prediction, but many high-performing models remain difficult to interrogate in clinically meaningful terms. This creates an interpretability paradox: gains in predictive performance often coincide with reduced transparency, while widely used post-hoc explanations can be persuasive without providing reliable evidence of model reasoning. Here, we present a critical narrative review and position argument, supported by a semi-systemat

The Artificial Intelligence Safety Measures Act requires developers of the most advanced AI models to yield to new levels of state oversight.

Attackers wasted little time targeting the latest memory disclosure flaw in Citrix's NetScaler products, after researchers published a proof-of-concept exploit (PoC).

How to apply effective governance to harness the benefits of A.I. and mitigate its risks ...

How to apply effective governance to harness the benefits of A.I. and mitigate its risks ...

Union und SPD wollen zur Politik hinter verschlossenen Türen zurückkehren. – Gemeinfrei-ähnlich freigegeben durch unsplash.com: Masaaki Komori Die schwarz-rote Koalition behauptet, geplante Einschnitte bei der Transparenz würden der Sicherheit dienen und die Nutzung des IFG erleichtern. Die Informationsfreiheitsbeauftragten von Bund und Ländern widersprechen vehement und warnen: Die Pläne würden Deutschland zurück in die Zeit des „verschlossenen Obrigkeitswissens“ katapultieren.

Novel category discovery aims to identify unseen classes from unlabeled data by transferring knowledge from labeled categories, but most existing methods perform discovery in opaque latent feature spaces. As a result, they may separate novel categories accurately while providing little insight into what semantic evidence defines each discovered group. We propose xNCD, an explainable novel category discovery framework that performs both representation-based discovery and pseudo-label assignment d

Ever wondered who gets to shape the historical record? Curious about questions of power, truth, and accountability? Want to know what a career in archives, records management, or information ...

Bangladesh has an estimated 1.17 mental-health professionals per 100,000 population and only six child psychiatrists nationwide. No Bengali-language, culturally adapted tool exists for early screening of abuse-related psychological trauma in children. We present ShishuRaksha AI, a decision-support (not diagnostic) framework that fuses four screening modalities: validated questionnaires (SDQ, CPSS), Bengali narrative text, House-Tree-Person (HTP) drawing features, and facial affect. The fusion is

In chest X-ray (CXR) classification, acceptable ranking performance can still leave rare-positive patients below threshold, especially within subgroups. We study this pre-deployment fairness problem as an audit question: after a long-tailed multi-label CXR model is converted from scores into decisions, who is missed? Across VinDr-CXR and MIMIC-CXR/CXR-LT, we use a diagnostic ladder to separate class-level long-tail losses, subgroup-aware weighting, group robustness, and threshold selection. On V

Reinforcement Learning (RL) has emerged as a powerful paradigm for adaptive traffic signal control. However, in safety-critical infrastructure like traffic control, the opaque, black-box nature of deep RL models poses challenges for transportation agency acceptance, regulatory compliance, operational trust, troubleshooting, and fine-tuning. To bridge this gap between high-performance optimization and human-comprehensible interpretability, this effort introduces a novel, explainable entity centri

Accurate and efficient dataset valuation is essential for enabling fair and transparent data marketplaces, especially when multiple contributors provide data for training multi-task models. Most existing valuation methods, however, are limited to single-task settings, overlooking scenarios where a buyer aims to optimize performance across multiple downstream tasks. Moreover, traditional valuation approaches, such as Shapley-based or retraining-based methods, are computationally expensive and poo