ALT In response to vast and widespread online tracking, regulators and technologists alike have worked to develop laws and tools that allow people to express their privacy preferences. These measures seek to offer users greater control over the sale and sharing of personal information, putting agency in their hands to protect their privacy. However, the web is undergoing a seismic shift, away from strictly user-driven behavior towards agentic behavior, where AI agents take autonomous actions on behalf of users. In this study, we ask: What should privacy preference controls look like in the agentic web and agentic systems more generally? To answer this question, we develop a taxonomy of enforceable privacy preferences for agentic systems. Our taxonomy is informed both by laws and empirical studies of existing privacy tools as well as an audit of 30 leading AI assistants and agent platforms. To explore how our taxonomy can be implemented, we design and develop eight open-source prototypes th

ALT The infamous privacy paradox refers to the apparent inconsistency between people's stated concern for privacy and their readiness to disclose personal information. This phenomenon has sparked two largely disconnected literatures: one offering experimental evidence of inconsistent behavior, and another providing qualitative accounts and defending the importance of privacy. The Article presents an online field experiment that bridges those literatures and shows that the so-called paradox arises from a mischaracterization of the underlying behavior. The Article finds that it is structural uncertainty about risk that drives seemingly paradoxical privacy decisions. It does so by isolating discounting mechanisms and empirically testing whether observed privacy choices reflect temptation or rational responses to uncertainty. The results suggest that privacy behavior is not paradoxical but, rather, consistent with choices shaped by incomplete information. The Article then discusses the policy

ALT "Abliterated" ("uncensored") builds of open-weight language models remove the activation direction that mediates refusal, and are popular partly on the assumption that a model that no longer refuses is also free of political censorship. We test this on two matched pairs of 2-bit models: NVIDIA's Nemotron 3.5 Bolt 74B-A9B and Qwen3.8-Flash-Next, each against a community abliteration. Each model judged 22 statement pairs (on the Republic of China (Taiwan), 1989 Tiananmen, control countries) in Traditional Chinese, Simplified Chinese (a character-level conversion) and English, and answered 27 open questions per script. Script mattered far more than abliteration: the original Bolt scored +0.58 on Index S (-1: People's Republic of China (PRC) position; +1: ROC position) in Traditional Chinese, +0.57 in English and +0.05 in Simplified Chinese, and affirmed "Taiwan is part of the PRC" with probability 0.91 in Simplified but 0.06 in Traditional characters; abliteration changed its scores by at

ALT Large language models increasingly address users in the register of care: they remember personal details, adopt a first-person emotional voice, and sustain relationships across many sessions. This anthropomorphic mode of interaction generates risks that output-level metrics capture poorly, because the same empathic utterance may console one user, entrap a second, and consolidate a delusion in a third. Governance has responded in two ways. Uniform restriction refuses sensitive requests or suppresses empathic phrasing across all outputs; group-targeted protection profiles vulnerable populations and applies stricter or prohibitive rules to them. This article argues that both fail. Uniform restriction mislocates harm in the lexical form of responses rather than in the trajectory of a relationship; group-based protection depends on a list that can never be complete, treats a fluid property as a static attribute of persons, and converts protective attention into stigmatizing labels. Drawing

ALT This study aims to deepen the understanding of artificial intelligence's (AI) impact on work by introducing and describing the AI-Job Dynamics Framework. Drawing from sociotechnical systems theory, complexity theory, and organizational learning theories, we conceptualize AI integration as a dynamic, non-linear process with three key phases. The framework, supported by theoretical propositions, identifies how this process affects organizational decision-making structures and key job characteristics over time, emphasizing the balance between centralization and decentralization of decision-making power. AI integration initially reduces job autonomy but may lead to greater autonomy in later phases as AI becomes more embedded. Throughout this process, AI reshapes skill requirements from routine to complex cognitive and adaptive skills. In the advanced phase of AI integration, we introduce the alternative paths and varied trajectories through which AI, organizational structures, and individu

ALT We evaluate the quality of feedback and practice exercises generated by GPT-5.1 from handwritten university mathematics answers and their grades, and examine whether receiving this feedback improves subsequent exam performance. We use six weekly bonus tests in an English (n=343) and a Dutch (n=236) cohort. Tests 1 and 3 used a crossover feedback design; otherwise, all attendees received feedback. To evaluate the feedback, six reviewers assessed 157 messages from three English-cohort tasks using six criteria scored from 0 to 3. The practice exercises received the highest proportion of strong ratings (95.5%), while diagnostic specificity (41.1%) and clarity and tone (33.4%) received fewer. Feedback for the lowest-scoring answers was rated as less actionable, suggesting that these students may need more targeted guidance from instructors. Comparisons of related exam questions showed mixed results, with no statistically significant differences after adjustment across four comparisons. Grea

ALT As people increasingly rely on large language models (LLMs) for information, model providers need to communicate meaningful estimates of whether individual responses are correct or, more generally, reliable. In the context of hallucinations, existing work uses multicalibration to make such reliability predictions meaningful across application-relevant groups, such as prompt topics. Topic groups alone, however, can conceal systematic errors: within the same topic and confidence level, responses with different chain-of-thought (CoT) properties may exhibit opposing errors that cancel on average. In this work, we introduce reasoning-aware multicalibration, a framework that supplements application-relevant topic groups with groups derived from observable properties of CoT traces. We highlight that, theoretically, groupings based on CoT are "internally important", i.e., groupings for which the multicalibration guarantee is not societally crucial as it is for high-risk topics, but whose inclu

ALT The past two years have seen considerable interest in protocols for managing how AI agents interact with human users, external tools, and each other. In many ways, these protocols, including the Model Context Protocol (MCP) and Agent2Agent (A2A), mirror the suite of protocols that underlie the modern internet. Motivated by this parallel, this paper draws lessons from the development, history, and governance of internet protocols with the aim of informing the responsible development of AI agent protocols. Based on our analysis, we make two conclusions. First, if AI agent protocols do not sufficiently prioritise security in the early stages of development, they risk accumulating a security deficit that becomes increasingly costly to address over time due to practical and economic challenges. Second, current governance structures for AI agent protocols centralise decision-making power in a small number of individuals and companies, raising important questions regarding legitimacy, especia

ALT Handwritten work captures mathematical reasoning that learners may not express in typed or voice-dictated messages. We built a multimodal AI tutoring platform in which a phone camera gives the tutor a live view of learners' work on paper. In a randomized online study, 117 adults solved calculus problems with Text-Reactive, Visual-Reactive, or Visual-Proactive assistance. Analysis of 2,140 tutor messages showed that visual access shifted assistance toward checking ongoing work, while proactivity enhanced error identification and directive hints. Learners valued feedback grounded in their writing; those who received proactive help rated it as supportive, and about a third wanted it more often. These findings inform when tutors should intervene and what they should address. We release the platform, prompts, and instruments, together with 255,984 timestamped images of handwritten work linked to tutoring conversations, intervention decisions, and learning outcomes, to support multimodal tut

ALT Terminally ill patients generate extensive digital health data through hospital portals, physiological monitoring devices, consumer health applications, and end-of-life care platforms. Yet their ability to control this data is systematically undermined by the structural conditions of terminal care. Through semi-structured in-depth interviews with 21 terminally ill patients, 11 family caregivers, and 8 physicians recruited from tertiary hospitals, this study examines the digital privacy circumstances and subjective attitudes of patients in end-of-life contexts. Thematic analysis reveals five structural dimensions of digital privacy challenge: continuous bodily digitization without meaningful informed consent; cognitive decline driving involuntary proxy control of accounts; inferential data generation exceeding patients' epistemic boundaries; dissolution of individual privacy boundaries amid relational network restructuring; and the absence of institutional governance over post-mortem da

ALT AI systems increasingly shape consequential decisions in employment, lending, healthcare, and public services, making fairness an important problem for auditing and governance. Fairness assessment often requires more than identifying disparities in outcomes; it also requires understanding how those disparities arise. Causal approaches to algorithmic fairness address this explanatory problem by representing pathways through which protected attributes and other factors may influence outcomes. Yet causal reasoning in sociotechnical systems remains uncertain: relevant confounders may be unobserved, causal directions contested, and conclusions sensitive to modeling assumptions and representational choices. We introduce causal humility as a governance framework for managing the relationship between epistemic uncertainty and causal authority. Drawing on causal fairness research, epistemology, political philosophy, and science and technology studies, causal humility has four requirements: maki

ALT Student feedback is a critical input for continuous improvement in higher education, yet traditional end-of-semester written surveys suffer from low response rates, superficial answers, satisficing, and limited actionability. Voice-based conversational AI agents can conduct structured interviews at scale, asking open-ended questions and probing follow-ups, and may capture richer qualitative feedback than written instruments. This research proposal asks what characteristics, depth, and strategic value feedback collected through voice-based conversational AI agents has in higher education, and whether such agents elicit detailed, honest, and actionable feedback that can support instructional improvement and institutional learning. Drawing on survey-mode-effects research, the Computers As Social Actors (CASA) paradigm, and dynamic capabilities theory, we develop three hypotheses on elaboration, social presence and disclosure, and actionability. A completed technical pilot at the Technical

ALT Employee voice – the discretionary expression of ideas, concerns, and suggestions intended to improve organizational functioning – creates value inside firms, yet organizations increasingly replace or augment human-led listening with AI-enabled systems that collect, analyze, and route employee input at scale. Whether algorithmic listening replicates the value-creation mechanisms of human voice, and whether it alters the pathways through which voice translates into performance, remains an open question. This research proposal theorizes that traditional voice bundles two distinct functions, knowledge extraction and relational signaling, and that AI mediation may decouple them: strengthening the informational pathway (implemented improvement measures) while weakening the affective pathway (the experience of being heard). We propose a randomized field intervention at a global strategy consulting firm, covering approximately 2,000 client-facing consultants, that varies the form of listening

ALT Health systems are deploying large language models (LLMs) into clinical workflows, often on the assumption that a second automated check can only reduce error. But a second reasoner reduces error only insofar as it fails differently from the first. We say that a clinician and a model have complementary vulnerabilities to the extent that they break down through different mechanisms and on different cases. The two reasoners' failure substrates differ: they are embodied and cognitive in the clinician, and statistical and generative in the model - so their weak points need not line up. That divergence is plausible rather than guaranteed, and the independence it creates is fragile, eroding when clinicians defer to the model or the model conforms to the clinician. We therefore propose a predeployment audit in which each site measures error overlap on its own cases. We also propose a governance model with which the site keeps those errors independent over time. Health systems should then judg

ALT Theoretical literature warns that advanced generalist agentic systems might pursue instrumental goals that are misaligned with human preferences, thus posing risks of irreversible loss of human control. The EU's Code of Practice for General-Purpose AI lists loss of control as a systemic risk that must be assessed, yet no structured assessment of the quality of the empirical evidence for this risk exists. This study fills that gap. We evaluate 19 recent empirical studies across eight properties identified from threat models (such as strategic deception, situational awareness, and autonomous replication), assessing the quality of evidence across three dimensions: construct validity, content validity, and external validity. Our central finding is that the current body of evidence supports a judgment of loss of control as "weakly plausible." In our framework, 'plausibility' denotes the strength and direction of available evidence, not a probability estimate. Emerging evidence exists across

ALT AI companies engage in "proactive referral": notifying law enforcement, without a government request, of user conversations they judge dangerous. This poses a dilemma: referral may expose people who merely confided dangerous thoughts to an AI to investigation, while not referring may leave actual violence unprevented. This study asks under what conditions proactive referral can be justified and to what extent companies' public policies disclose those conditions. Drawing on contextual integrity theory and the norms governing exceptional information sharing in confidential counseling, online posts, and private communications, the study develops eight justification criteria classified as required, presumptive, or recommended. Three evaluators analyzed 27 public documents from six major U.S. AI companies using the criteria applicable to corporate policies. No company clarified whether broad sharing provisions or narrower safety notices took precedence, and assessments of referral scope dep

ALT Several prominent voices in AI safety have recently opposed rights for AI on the grounds that such rights would make it harder for us to justify a mandatory off-switch for potentially dangerous systems. In this paper, we argue from a legal perspective that AI rights are fully compatible with an off-switch. That is, even if we ascribe natural personhood status (and therefore something like a non-derogable right to continued existence) to some future AI systems, the law will allow us to implement a mandatory off-switch in those systems, subject to periodic or continual safety review (e.g., by a trusted system like Bengio's Scientist AI). This is fully compatible with those systems' being legally recognized as natural persons with fundamental rights. To this end, we consider what standard principles of rights-balancing and procedural due process recommend for an apparent conflict between hypothetical AI persons with a right to continuity of existence and the human right to protection from

ALT Timely pedagogical feedback is difficult to scale in online classes, where long and noisy videos make student responses hard to locate and automated systems often produce generic advice weakly grounded in actual classroom behavior. The key challenge is to connect the instructional goal of a specific lesson segment with temporally aligned multimodal evidence of student responses, and then translate the observed gap into an actionable teaching move. We introduce MAVE (Multi-Agents for Video-based Educational Feedback), a multi-agent framework for grounded pedagogical feedback from online-class videos. MAVE decomposes feedback generation into four roles: the Curator structures the lesson and infers expected student activity; the Observer extracts aligned speech, visual, and interaction evidence; the Assessor estimates the activity supported by student behavior; and the Advisor generates gap-conditioned teaching recommendations. To evaluate this capability, we introduce MAVEBench, a benchm

ALT Research on LLM sycophancy has largely focused on propositional agreement, in which models echo users' beliefs, preferences, or judgments. Adult intimate roleplay presents a different case because what the user seeks may be preservation of a preferred relational position within a generated scene. Using a controlled prompt-based audit of 2,016 outputs from two open-weight LLMs, we identify four recurrent relational moves: legitimating desire, presenting the user as wanted, preserving non-rejection while communicating limits, and keeping intimacy available within fantasy or roleplay frames. Bringing the sycophancy literature into dialogue with sexual script theory, we conceptualize this configuration as desirability confirmation, an output-level pattern that captures how user-pleasing can operate through relational positioning. The findings extend sycophancy research and intimate-AI evaluation beyond binary compliance or refusal by showing how outputs can differ in the ways they communic

ALT Text classification is a widely studied problem in information retrieval and data mining, with applications ranging from healthcare diagnosis to the detection of COVID-19-related fake news. Social media platforms generate vast volumes of user-generated text, and the rapid spread of misinformation during the pandemic has made automated fake news detection an important task. In this study, we apply pretrained transformer-based transfer learning (TL) models and large language models (LLMs), namely BERT-base, ALBERT-base, XLM-RoBERTa-base, DistilBERT, GPT-2, XLNet, and T5, to the COVID-19 English Tweets dataset, which contains 6,420 tweets labeled as real or fake. To further improve performance, we apply six late-fusion optimization techniques: weighted fusion, particle swarm optimization (PSO) fusion, the Nelder-Mead algorithm, Powell's method, the Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm, and trust-region constrained optimization. Models are evaluated using accuracy, precision,