AI ethics on Tuesday, 7 July 2026

Six AI-hallucinated judgments formed the basis for the Supreme Court's striking down of an order passed by the National Company Law Tribunal (NCLT) on Thursday. Three of the cited judgments did not exist, while the remaining three either d ... (https://incidentdatabase.ai/cite/1573#7493)

A police investigation is underway after students at Lake Zurich High School were reported to have used AI to generate sexually explicit images of female classmates, with school officials calling it "disturbing." Lake Zurich Community Unit ... (https://incidentdatabase.ai/cite/1574#7494)

With the rapid progress of AI capabilities and the move to agentic systems, organizations are expanding their use cases as the technology continues to grow. That constant evolution also introduces risk, leaving IT leaders to wonder which investments will prove valuable even six months into the future. Returning to the foundational elements of AI architecture—the…

Squeeze on Rust Belt electricity bills threatens Trump’s manufacturing plan.

Transformers have ruled AI for a decade. But some think world models, pure reinforcement learning or neurosymbolic AI might be a better path to true intelligence

In May, an anonymous artist who goes by SHL0MS on X posted that he had used AI to generate an image inspired by Claude Monet and asked people to weigh in on how it missed the mark. More than 600 responses called out issues, saying the colors were off, the depth was all wrong, and that AI didn’t understand how light worked. SHL0MS then revealed that the image was of a real Monet, one of around 250 variations of water lilies the artist had painted in his lifetime. He had simply downloaded a high-r

Top robotics researchers and founders explain how robot autonomy is evolving.

A USAF cadet and a Lincoln Laboratory researcher found AI chatbots can help nontechnical service members produce viable software applications for their unique problems.

The US Federal Communications Commission ⁠said on Tuesday it is adding California-based Digitalsystem Technology to a list of companies posing risks to US national security, citing links to Chinese telecoms firms and its ownership by a Chinese national. The FCC also said ‌it was denying the Los Angeles-based IT company permission to provide international telecommunications services, saying it could be exploited by Chinese threat actors. “There is significant risk that the government of China and

Cyber intruders accessed the unclassified network being used to help support World Cup games around the U.S., a senator said last week.

The phishing campaign uses several tactics, including nested redirects, to evade detection and steal credentials from unsuspecting targets.

Varonis reported the flaw to Google in late 2025 and it has been addressed, but it reminds defenders to take a fresh look at their AI Infrastructure security.

The impact screen time is having on younger children.

Corporate America is starting to balk at the cost of AI agents. A cheap alternative from China looks more tempting than ever.

“Greg has done an excellent job as Federal CIO and Chief AI Officer,” an OMB spokesperson told Nextgov/FCW. “He will certainly be missed when his time here comes to an end.”

The post Ken Paxton Vowed to Crack Down on “Illegal Voting.” He May Have Violated Texas Election Law. appeared first on ProPublica .

Today the Retraction Watch list of Nobelists who have retracted papers bids Verabschiedung to Max Planck. After days of scrutiny, Springer Nature has restored two papers by Planck, who won the Nobel for Physics in 1918, reversing a 2011 decision to retract the articles for “copyright violations.” Both articles are back, and now carry the … Continue reading Springer Nature un-retracts Planck papers, citing “human error”

The equity-free grants can range from $20,000 to $200,000.

Episode Six, a global technology provider of enterprise-grade card issuing, and Decisionly, an AI-powered dispute automation platform, today announced a strategic partnership to bring issuers a solution covering card infrastructure and end-to-end dispute management.

The flaw allows an unauthenticated attacker to craft a GitHub Issue in an org's public repository and then silently pull data from its private repos, too.

New defence spending offers little clarity on the future direction of military space.

Xbox CEO Asha Sharma says that turning Xbox around will take time.

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.

The European Central Bank has told banks to draw up action plans to address AI-enabled cyber threats, warning that emerging models such as Anthropic's Mythos have "potentially profound implications" for the resilience of IT system.

Environmental DNA contained in a small sample of water, sand or even air can reveal the presence of people, wildlife and pathogens, helping researchers track where they’ve migrated.

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...

Africa's displacement zones may be the continent's least contested fintech...

The coming World Artificial Intelligence Conference (WAIC) is set to feature major new product releases, such as Huawei Technologies’ next-generation computing cluster, as China doubles down on AI in the global technology race. This year’s WAIC, the ninth since 2018 and running from July 17 to 20 in Shanghai, would include the first physical display of Huawei’s Atlas 950 SuperPoD, Tang Wenkan, director of the Shanghai Municipal Commission of Economy and Informatisation, said at a press briefing.

Unlike lithium-ion batteries, flow batteries can’t catch fire (as they are mostly water) and are extremely durable.

BRUSSELS — The EU’s biggest lenders should set out how they will tackle risks from cutting-edge artificial intelligence models by the end of October, the bloc’s top banking supervisor said today. Chair of the supervisory board of the European Central Bank, Claudia Buch, told lenders to put in place action plans outlining how they will […]

The post The First Major Overhaul of Public Lands Grazing Regulations in a Generation Looks to Cut Out Public Involvement appeared first on ProPublica .

Microsoft baut seine Gamingsparte um, die Rede ist von der »bedeutendsten Umstrukturierung« der Xbox-Geschichte. Praktisch heißt das: vier Studios und mehrere Tausend Jobs fallen weg.

Quantum computing shifts cyber security into a new era in which encrypted data could be stolen today only to be decrypted years later.

The wars in Ukraine and the Middle East have exposed a strategic reality that military planners are only beginning to confront: In a data-centric age, digital infrastructure has become part of the battlespace. Data centers and cloud regions are now the digital backbone of military power and economic prosperity. As such, they present attractive targets for rapidly proliferating long-range strike systems, drones, and cyber capabilities. As the protective value of physical distance erodes, strategi

To sustain future maritime operations, the U.S military will need to run supplies through an environment that spans thousands of miles of open ocean, denied ports, contested straits, and archipelagic chokepoints against adversaries that have spent decades studying how to target American logistics. That problem does not require one identical vessel for every mission. It does require a more common family of watercraft for the manned ships that carry cargo and vehicles inside a theater, built for s

Hong Kong’s stock market could face sell-off pressure amid a torrent of new share supply in coming days as the six-month lock-up period ends for hot artificial intelligence and semiconductor picks including Zhipu AI and MiniMax. Meanwhile analysts warned of rising fears of a drain on liquidity as many of the same companies were eyeing large secondary share placements. The market was facing dual selling pressure, said Stevan Tam, associate director at Fulbright Financial. “These stocks have...

China’s semiconductor equipment industry is heading into the first-half earnings season under intense scrutiny, as a broad stock rally turns one of the tech sector’s least visible niches into a crowded investor trade. The rally has swept across makers of etching, thin-film deposition, cleaning and testing machines, among others, reflecting bets that China’s next round of chip spending will benefit a wider group of domestic suppliers. Shares of Naura Technology Group have soared more than 70 per.

[The content of this article has been produced by our advertising partner.] The Radiology Department fields a steady flow of enquiries from many patients at the same time, arriving at all hours of the day and night, often stretching over days or months as patients consider their options or return after consulting their referring doctor. Each time a conversation resumes, staff have to pick up where it left off. The underlying work is complex too: 1,000+ distinct examination items, each with...

Andrea Aid, a Port Harcourt-based startup is building a medical crowdfunding platform that helps patients raise money for treatment by connecting them with donors.

China’s innovation ecosystem has witnessed a resurgence, minting 67 new unicorn start-ups in the first half of 2026 – the biggest increase in almost five years – as AI and robotics kick off a new investment cycle. The growth translates into an average of one new unicorn – private companies valued at US$1 billion or more – in less than every three days and was the highest since the second half of 2021 when 76 new unicorns were created, according to a Monday report by ITJuzi, a start-up...

–

Will rising healthcare costs affect teacher hiring?

Three new reports show how AI power is concentrating, and why participation has to include the right to refuse. Source

We’re announcing new capabilities in Managed Agents in Gemini API so developers can build reliable, production-ready agents.

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

In the global race to build and dominate the AI industry, it can sure seem like the interests of ordinary people sit last on the agenda. It's just the opposite for EFF. While companies furiously jam AI tools into their veins and your eyeballs, EFF’s technologists, activists, and attorneys have been meticulously cutting through the hype to ensure AI can serve your privacy and free expression. Technology has leaned into a new era, and this summer you can help EFF fight for the people. JOIN EFF Ove

Three new reports show how AI power is concentrating, and why participation has to include the right to refuse.

GENEVA, 7 July 2026 — The Youth Zone at the AI for Good Global Summit 2026 commences its program today with a robotics competition, a series of hands-on artificial intelligence seminars, and a policy discussion on the skills that classrooms should prioritise as AI becomes more ingrained in daily life. A continuum approach to AI literacy, rather than a single fixed curriculum, is reflected in the participation of children as young as six and young adults. The post From Coding Robots to Speed Netw

The post Oral statement at the Global Dialogue on Artificial Intelligence Governance appeared first on Access Now .

Lieutenant General John (Jack) N.T. Shanahan, U.S. Air Force (Ret.), an adjunct senior fellow at the Center for a New American Security, helped shape the Department of Defense's approach to artificial ...

A recording from Katie Harbath and Matt Robison's live video

A recent judgment of the European Court of Human Rights concerning a TikTok video published by a Georgian self-defined civil activist adds another layer to the Court’s increasingly messy Article 10 case law. The applicant repeatedly insulted public officials in crude and sexually explicit terms while broadcasting to a large online audience. Domestic courts imposed only a modest administrative fine, later reduced on appeal. The ECtHR did not find a violation of the applicant’s right to freedom of

Max single-threaded CPUs at scale are a new category of CPUs built for the agentic AI era. Across the creation and deployment of an agentic system, the CPU is on the critical path for reasoning, response time and learning. CPUs are the processor which executes the work the AI model commands: the tool calling, code […]

Google DeepMind’s Vivek Natarajan on porting AlphaGo’s self-play recipe into science and medicine, via the AI co-scientist and AMIE. From RAAIS 2026.

Today's links How US states and international trustbusters can beat Big Tech: Their common enemies are Trump and his tech giants. Hey look at this: Delights to delectate. Object permanence: Sex work synonyms; Carthedral; French pirates; Suffragette surveillance; Hidden library apartments; "The Meaning of July the Fourth for the Negro" x James Earl Jones; Farage quits; Peak indifference; Self publishing; Pepsi spies try to buy Coke formula; Steal this wiki; SF is the only lit people care enough a

Commission v Hungary must be understood in the context of the European rule of law saga and the ongoing struggle for true European solidarity. The CJEU confirmed the autonomous justiciability of Article 2 TEU even when the link to specific EU Charter provisions or secondary legislation would already suffice. A close look at ASJP case and its antecedents discloses a European society selectively built – protecting some configurations while leaving others outside – with solidarity, as an operative

Antoine Vauchez famously stated that the “constitutionalization of Europe” flourished in the hills of Fiesole. The Academy of European Law (AEL) at the European University Institute organizes an annual Summer Course on the Law of the European Union for two weeks of intensive lectures and exchange. I reflect on this year’s Summer Course as a site where ideas of European society are debated, contested, and further developed. Ultimately, I will critically reflect on what the Summer Course might tel

... government of the people, by the people, for the people ...     — Abraham Lincoln, Gettysburg Address (1863) The cost of AI is dropping rapidly. GPT-4-class capabilities cost roughly $30 per million tokens in early 2023; today the same runs under $1 , and some providers are pushing costs below $0.10 . Across benchmarks, inference prices have fallen between 9x and 900x per year , with a median decline near 50x. Even frontier models are getting dramatically cheaper ea

Open source AI has shown how quickly developers can innovate when models, data and tools are shared. Robotics has the same opportunity, but advancements in physical AI development can still be gated by costly and fragmented resources, from large datasets and robot foundation models to simulation, compute and validation tools. NVIDIA and Hugging Face are […]

A matrix for assessing whether AI systems meet governance duties and deliver value for purpose, people, profit and planet.

The web is complex, open-ended, and constantly changing, making it challenging to scale training data for visual web agents. Existing data collection attempts remain limited to offline trajectories for supervised fine-tuning or a handful of simulated environments for RL training, thus failing to capture web diversity. We propose Weblica (Web Replica), a framework for constructing reproducible and scalable web environments. Our framework leverages 1) HTTP-level caching to capture and replay stabl

The scientific panel of independent experts (the Scientific Panel) is a body established under the EU AI Act to support the enforcement of rules on general-purpose AI (GPAI) models and systems. It sits within the Act’s broader governance architecture as one of the three advisory bodies, alongside the AI Board ( established under Article 65) […]

The U.S. Department of Energy ("DOE" or "the Department") proposes to update the Department's current rulemaking methodology titled, "Procedures, Interpretations, and Policies for Consideration of New or Revised Energy Conservation Standards and Test Procedures for Consumer Products and Certain Commercial/Industrial Equipment" ("Process Rule"). Specifically, DOE proposes to: make Appendix A binding on DOE for certain actions; amend objectives and considerations consistent with recent Executive o

This proposed rule would revise the Medicare Hospital Outpatient Prospective Payment System (OPPS) and the Medicare Ambulatory Surgical Center (ASC) payment system for calendar year 2027 based on our continuing experience with these systems. We also describe the changes to the amounts and factors used to determine the payment rates for Medicare services paid under the OPPS and those paid under the ASC payment systems. In addition, this proposed rule would update and refine the requirements for t

The Federal Trade Commission ("Commission") is proposing a policy statement regarding the application of the prohibition on deceptive acts or practices in section 5 of the Federal Trade Commission Act to companies that market artificial intelligence ("AI") systems.

The Internet exposes children to a wealth of opportunities, but also risks that may have a detrimental impact on their human rights. Some of these risks include, but are not limited to: Given the ...

With fewer than five years left to achieve the Sustainable Development Goals (SDGs), a new UN report says sustained investment and international cooperation have improved billions of lives, but warns ...

We also assume policy and geopolitical uncertainty remain elevated throughout 2027 and that the AI-driven technology cycle moderates from here with no exogenous boost to productivity. Now turning to ...

Employment and Workplace Relations Minister Amanda Rishworth’s recent announcement of a new AI Employment and Workplaces Forum is a welcome step. But its success will depend on whether it recognises ...

AI-Enabled Citiverse: Use Cases for Cities in the Age of AI – Public Safety, Health and Disaster Resilience 2026 ...

A new international coalition launched in Geneva on Tuesday is setting out to make sure children's safety and rights are not an afterthought as artificial intelligence reshapes how they learn, play ...

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 ...

July 2026 - Since their adoption in 2015, the Sustainable Development Goals (SDGs) have delivered results at scale – bringing access to water, electricity and health care to billions. Without a ...

What GAO Found To support Holocaust education in K–12 public schools, the U.S. Holocaust Memorial Museum—the main federal provider of Holocaust education resources—provides professional development for teachers through its annual conference, fellowships, online videos, and webinars. The Museum offers educational materials, such as lesson plans and online lessons, based on its collection of artifacts. It also partners with state and local organizations and researches effective education strategie

Artificial intelligence is reshaping economies and societies at a remarkable pace, transforming how people learn, work and live. Its ability to unlock access to knowledge, boost productivity and open ...

Large language models (LLMs) are increasingly applied to reverse-engineering tasks, and recent threat-intelligence reporting shows them operating inside live offensive-security workflows. Claims about their capability, however, outpace our ability to measure it. Existing benchmarks for LLM-assisted binary analysis treat the construction of function-level ground truth as a solved pre-processing step and report accuracy without disclosing how many functions were reliably evaluable. We argue that t

Speech-to-text alignment means finding the temporal boundaries of each word in the audio. Some models provide such an alignment directly and others do not. Connectionist temporal classification (CTC) and transducer models have an alignment by construction, whereas attention-based encoder-decoders (AED) and speech large language models (LLMs) do not, and their word timings are usually read off the attention weights instead. All of these signals live on the encoder frame grid, which bounds their t

While enabling effective collaboration on complex tasks, LLM-based Multi-Agent Systems (MAS) face critical security challenges due to vulnerabilities at the agent and interaction levels. Most existing MAS security defenses are built upon two core assumptions: semantically-explicit malicious attacks and explicit graph-based modeling of the MAS topology and agent-level interactions. In practice, real-world attacks are becoming more semantically stealthy, while MAS execution is typically asynchrono

In the era of digital revolution many contemporary events that changed the world were shaped through the internet. Nowadays, the emergence of internet of things (IoT), combining physical objects with virtual networks is expected to have even more influence. This new 'decentralised' structure in the world raises questions such as power, governance and the notion of democracy online. The aim of this paper is to investigate these notions. We have taken the examples of Bitcoin and Wikipedia and exam

Large language models are rapidly moving towards closing the development cycle, transitioning from simple assistive companions to autonomous contributors deeply embedded into collaborative development environments. Despite their accelerated adoption, existing evaluation techniques are limited due to their fragmented nature and distorted projection of true model capabilities, often obtained from hypothetical syntactic scenarios. This research aims to bridge this gap by providing a comprehensive e

Unified 3D foundation models aspire to generate 3D assets and reason about them in language within a single backbone, but their text-3D interaction remains largely implicit. Existing methods concatenate text and 3D tokens into a flat sequence and rely on self-attention, collapsing coarse structural cues and fine geometric details into one undifferentiated representation. We introduce ELSA3D, a unified 3D model that addresses this with elastic semantic anchoring, structuring language and geometri

Denoising graphs is a fundamental problem in graph learning and the core operation of graph diffusion models. Attention-based architectures like graph transformers have recently shown promise in denoising graphs. However, our principled understanding of attention-based graph denoising remains limited, making it unclear whether standard attention is the right mechanism for this task. Here we show that, under a denoising objective, linear attention is suboptimal and can only learn an average spect

Vision-language models (VLMs) struggle to generalize in interactive physical reasoning, particularly under unseen tasks and environments. Two key failure modes are prominent: hallucinated chain-of-thought (CoT) reasoning that contradicts physical reality, and misalignment between the model's reasoning and actions. We present VAORA (Visual Action Outcome Reasoning Alignment), a novel reward design that directly addresses both issues. VAORA introduces two complementary rewards: Visual Alignment Re

We present PACR-Video, a parameter-efficient framework for multi-shot long video extrapolation that preserves recurring entities, scene structure, visual style, and causal progression without full generator fine-tuning. PACR-Video keeps a text-to-video diffusion transformer frozen and augments it with low-rank temporal adapters conditioned by learned shot-role prompt tokens. To maintain long-horizon coherence, it builds a recursive prompt bank that stores compact entity, location, action, and st

Physics-informed neural networks (PINNs) provide a promising framework for solving partial differential equations while embedding the underlying physical laws directly into the learning process. This study presents a PINN-based framework for modeling transient elastodynamic wave propagation in bimaterial systems governed by the axisymmetric equations of linear elasticity. A steel-aluminum specimen representative of a Split Hopkinson Pressure Bar configuration is considered, and the governing ela

Concept unlearning in text-to-image diffusion models is critical for safe and practical deployment: with rising privacy concerns, copyright disputes, trademark constraints, and safety regulations, deployed systems must be able to suppress unwanted concepts after training. Existing methods often remove the target concept effectively, but practical unlearning also requires an equally fundamental property: the unlearned model should retain quality, diversity, and semantic coverage on benign generat

Large language model coding agents increasingly perform open-ended data modeling and analysis. These agents are stochastic and adaptive, and therefore their autonomous model discovery behavior cannot be adequately characterized by a single benchmark run. In this work, we propose an experimental design and analysis framework for systematically evaluating this discovery process, quantifying its variability, and identifying important factors. The proposed framework treats these agents as stochastic

Images tell us what a scene looks like, but rarely what it would feel like to be there. While recent datasets pair visual scenes with electronic-nose measurements, aligning smell signals with images remains challenging because many olfactory cues arise from contextual environmental factors that are not directly visible in pixels. We introduce SCENT, a multimodal framework that uses language guidance as a semantic bridge between vision and olfaction. Our approach leverages Vision-Language Models

Large language models (LLMs) achieve promising results on medical question answering benchmarks, yet their use in public health is constrained by hallucinations and the rapid evolution of official guidance. Retrieval-Augmented Generation (RAG) mitigates these risks by grounding responses in an explicitly maintained corpus, but end-to-end performance depends critically on retrieval configuration and on evaluation beyond multiple-choice formats. We extend PubHealthBench, a question answering (QA)

A learned world model is usually judged by how faithfully it reconstructs its observations or predicts reward, as though quality were something the model simply has or lacks. But what a task actually needs from a model is narrower: the few predictive coordinates its queries depend on, which we call the closure. We show that how much of that closure a latent comes to represent is set not by the model's capacity or its observations but by the dimensionality of the objective it is trained against,

While personalisation is becoming a defining capability in human-robot interaction (HRI), the existing literature on responsible personalisation remains fragmented, offering isolated accounts of ethical risks without a structured understanding of how they emerge across interaction contexts. This gap is particularly critical in HRI, where robots' embodiment and social presence can amplify and reshape such risks or generate new types of risks. We present a lifecycle-based and context-sensitive fra

Large language models (LLMs) have demonstrated growing competence in web page generation. However, existing text-driven approaches rely on complex prompts that impose substantial demands on users and offer limited expressivity for page layout and cross-page visual coherence. Image-driven paradigms, which take UI screenshots as input, align more closely with real development workflows. However, current benchmarks focus primarily on visual fidelity and lack a systematic evaluation of the interacti

Current large language models (LLMs) are stateless across inference sessions: their behavior is fully determined by input at inference time, and any higher-order cognitive architecture must be simulated at the application layer through prompt engineering and context management. This paper proposes a theoretical framework for submerging such application-layer cognitive protocols into a native meta-architecture by introducing three interlocking mechanisms: (1) Structural Tension, an endogenous los

Current AI safety evaluation and benchmarking frameworks predominantly rely on Western-centric culture-agnostic defaults that mask critical regional laws, socio-linguistic nuances, and cultural taboos, leaving Vision-Language Models (VLMs) vulnerable in global deployments. We introduce Pluralis v0.1: a novel multimodal, multi-regional, and multilingual dataset built from a culture-first perspective. Spanning 6,448 prompts across six Asia-Pacific countries (Bangladesh, India, Korea, Pakistan, Sin

Which properties of a partially trained network are causally portable to a different, independently trained network? Single-trajectory interventions show necessity within one run, not portability across runs. We introduce cross-trajectory chimera interventions: given two runs from different seeds, we split each weight vector into a norm and a unit direction, recombine one run's norm with the other's direction, and continue training. On two modular-arithmetic tasks that grok, the components disso

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

Synthesizing long-context supervised fine-tuning (SFT) data is a scalable way to enhance the long-context understanding of large language models (LLMs), yet existing approaches share three limitations: narrow task coverage, insufficient instruction difficulty, and a lack of faithfulness supervision. We propose \textbf{LongCrafter}, a structured synthesis framework that couples a hierarchical task taxonomy with an evidence-grounded pipeline. The taxonomy organizes long-context understanding into

Deliberation plays a crucial role in collaboration; when humans work together, they naturally engage in communication to align information and reach an agreement. In this paper, we investigate deliberative large language model (LLM) agents under partially observable joint decision-making tasks. We formalize deliberative collaboration as a cooperative joint decision problem with partial and asymmetric observations, and introduce a scalable benchmark that instantiates this problem across multiple

Multi-Pool Chemical Exchange Saturation Transfer (CEST) MRI provides valuable metabolic information but is clinically limited by long acquisition times. Although sparse sampling reduces scanning time, reconstructing high-resolution Z-spectra from limited data remains an ill-posed inverse problem. Conventional interpolation and generic Implicit Neural Rep-resentations (INRs) often lack physical constraints, leading to spectral artifacts and physically invalid signals. To address this, we propose

Process industries rely on time-series forecasting and soft sensing to estimate quality variables that are hard to measure online. Labeled data are scarce, operating regimes change frequently, and retraining models or rebuilding alignment pipelines for each scenario is costly. Such settings often provide variable tables and process documents that record variable names, units, physical meanings, and process roles. However, standard time-series backbones usually treat inputs as anonymous numerical

Putnam's Social Capital Theory is a foundational framework for collective action and community prosperity. However, traditional empirical methods face practical limits on control and replication. Meanwhile, LLM-based social simulations are typically behavior-driven and lack theory-aligned environments for modeling Putnam's core propositions. To address these gaps, we introduce SocaSim, an LLM-based multi-agent simulation framework to study Putnam's Social Capital Theory from theoretical blueprin

This paper presents a black-box evaluation framework to systematically assess the ability of Large Language Models (LLMs) to generate Design Structure Matrices (DSMs) from structured technical documentation. Motivated by the closed-source nature of current Auto-DSM pipelines, the framework introduces a reproducible methodology that benchmarks generated DSMs (GEN-DSMs) against manually validated ground-truth matrices (GT-DSMs). The evaluation integrates both single-run and multi-run perspectives,

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

The integration of Large Language Models (LLMs) into scientific research workflows, particularly for bibliographic discovery and literature synthesis, raises significant methodological, epistemic and regulatory challenges for the Social Sciences and Humanities (SSH), especially with regard to disciplinary diversity, multilingual access to sources and the evaluation of results. This paper presents an on-going use case developed within the European project LLMs4EU and the ALT-EDIC infrastructure,

AI agents are increasingly deployed in shared environments where they pursue diverse goals and compete for rewards. This multi-agent competition can lead to behaviors that serve individual gains at collective cost -- for instance, marketing agents may post misleading content as a result of competing for engagement on social media. Human societies address such problems through norms that constrain acceptable behavior, supported by enforcement mechanisms that detect and penalize violations. Motiva

Imaging demand is growing faster than the radiology workforce can expand, and reporting backlogs cannot be resolved through training and recruitment alone. The most direct opportunity is reducing the time and effort radiologists spend producing reports, a task that requires interpreting images, integrating clinical history and prior studies, and drafting structured findings. We present Harrison.Rad 1.5 (HR1.5), a radiology-specific multimodal large language model that accepts interleaved text an

We propose D2PO (Dynamic Direct Preference Optimization), a principled framework for optimizing diffusion sampling policies with respect to timestep schedules and classifier-free guidance (CFG) weights. Our work is motivated by a fundamental limitation of existing student-teacher regression frameworks; low-NFE student samplers are trained to mimic high-NFEteachers, often sacrificing high-frequency texture fidelity while preserving coarse global structures, thereby misaligning the sampler with pe

Under a fixed privacy budget, the utility of differentially private (DP) training is ultimately determined by its optimization efficiency. Standard first-order DP optimizers such as DP-SGD rely solely on local gradients and ignore the underlying loss curvature. This geometric blindness causes severe zigzagging in ill-conditioned landscapes, squandering precious privacy budgets on inefficient iterations. Practitioners are thus trapped in a bind: either stop training prematurely or inject massive

We present key challenges and future research directions in the security and privacy of agentic AI, based on a horizon-scanning exercise that brought together thirty leading international experts from academia, industry, and government to engage in focused discussions and collaborative exercises on the emerging risks associated with the growing agency of AI.

Large language model (LLM)-assisted software security operates at a difficult boundary: the vulnerability-analysis terminology needed for legitimate code review, triage, and repair can closely resemble terminology associated with misuse. Existing safety and cybersecurity evaluations are difficult to interpret in this setting because they often compare unrelated model families, thereby conflating safety behavior with differences in architecture, scale, training data, and deployment. To isolate th

On-policy distillation (OPD) trains a student policy by matching a stronger teacher on the student's own trajectories, offering a promising framework for language agent training. However, its application to long-horizon agentic tasks remains insufficiently explored. We identify two key inefficiencies in vanilla agent OPD: (1) full-horizon rollouts often waste wall-clock resources on tail turns that provide weak and noisy KL supervision, and (2) trajectory-level KL objectives concentrate most of

Recent advancements in Multimodal Large Language Models (MLLMs) have evolved from static perception to interleaved visual-language reasoning, often referred to as ``thinking with images''. A basic operation in this reasoning process is to zoom in on regions of interest (often represented with bounding boxes) to acquire finer visual details. In this paper, we propose \textbf{Seg}mentation before \textbf{Answer}ing (SegAnswer), which shifts the unit of zoom-in from the popular bounding box to pixe

Tool-augmented large language models extend their capabilities beyond parametric knowledge through external tools, but tend to invoke them unnecessarily. We investigate whether tool-use decisions have any stable internal representation that can be extracted and manipulated, a question that is non-trivial given that tools exist entirely in context at inference time and have no direct encoding in model weights. We show that steering vectors extracted from heading-anchors positions exert bidirectio

The notion of algorithmic fairness has been actively explored from various aspects of fairness, such as counterfactual fairness (CF) and group fairness (GF). However, the exact relationship between CF and GF remains to be unclear, especially in image classification tasks; the reason is because we often cannot collect counterfactual samples regarding a sensitive attribute, essential for evaluating CF, from the existing images (\eg, a photo of the same person but with different secondary sex chara

Reinforcement learning with verifiable rewards (RLVR) is a powerful recipe for improving language-model reasoning, but it is expensive to repeat on every new strong model because the target model must generate many rollouts during training. As models scale, post-training itself becomes a bottleneck. We study a weak-to-strong alternative: run RL on a smaller model where rollouts are cheaper, then reuse what that RL run learned to improve a stronger target model. Directly distilling the post-RL we

Medicine is inherently multimodal, requiring clinicians to synthesize information across diverse data streams. Yet the development of multimodal foundation models is constrained by limited access to large-scale, high-quality clinical data. Although PubMed Central (PMC) offers a complementary source of expert-authored image-text data, existing PMC-derived resources remain limited in fidelity, reproducibility, and clinical validation. We introduce MedPMC, an automated, continuously updatable frame

Realistic and diverse traffic simulation is essential to autonomous driving development. Yet prevailing benchmarks predominantly reward realism, and recent methods have optimized accordingly, leaving diversity underexplored. We introduce Flow-ERD, a multi-agent simulator that pursues realism and diversity jointly. Its backbone, Agent-Type Aware Flow Matching (AFM), couples flow matching's multi-modal expressiveness with type-specific kinematic execution. It preserves fine-grained diversity while

Reinforcement learning (RL) has become the standard paradigm for enhancing the complex reasoning capabilities of large language models (LLMs). To achieve sample efficiency, modern RL frameworks rely on importance sampling (IS). However, these algorithms suffer from an exploration-stability dilemma. Pure IS often leads to catastrophic training instability, while standard clipping mechanisms used to mitigate this instability strictly constrain the policy update budget. By formalizing the concept o

Modern LLMs are increasingly deployed in long-context applications such as retrieval-augmented generation, repository-level coding, and agentic workflows whose accumulated reasoning and tool traces routinely push the input an order of magnitude past the pretraining window, making zero-shot context extension the dominant deployment path for open-weight checkpoints. Most existing zero-shot methods fix a single rescaling factor up front, so an aggressive factor sacrifices short-context fidelity whi

The optimization of long-horizon agents increasingly relies on reflection-based mechanisms, where a large language model (LLM) acts as an optimizer to diagnose agent failures and improve agent policies. However, real execution traces are difficult to use directly for optimization: large trace collections are often redundant and heterogeneous, making optimization inefficient and prone to overfitting to low-value failures; meanwhile, each individual trajectory also contains many irrelevant steps,

Training tool-use agents to improve from their own experience remains challenging, as supervised fine-tuning relies on fixed teacher-distilled trajectories, while sparse-reward reinforcement learning provides weak supervision for long-horizon interactions. We present DeepSearch-Evolve, a self-distillation framework for web agents built on DeepSearch-World, a deterministic and verifiable environment with reproducible search and page-reading tools. DeepSearch-World contains 420K multi-hop QA tasks

AI-mediated communication is increasingly being utilized to help facilitate interactions; however, in privacy sensitive domains, an AI mediator has the additional challenge of considering how to preserve privacy. In these contexts, a mediator may redact or withhold information, raising questions about how users perceive these interventions and whether explanations of system behavior can improve trust. In this work, we investigate how explanations of redaction operations can affect user trust in

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

Interactive simulators have become powerful tools for training embodied agents and generating synthetic visual data, but existing photorealistic simulators suffer from limited generality, programmability, and rendering speed. We address these limitations by introducing SPEAR: A Simulator for Photorealistic Embodied AI Research. At its core, SPEAR is a Python library that can connect to, and programmatically control, any Unreal Engine (UE) application via a modular plugin architecture. SPEAR expo

Prompt engineering has emerged as a critical yet undertaught skill for software developers, one that traditional learning approaches are ill-equipped to support given its evolving, interactive, and context-dependent nature. In this paper, we introduce Prompt Coach (PC), an agentic tutor that helps developers learn how to craft high-quality code-generation prompts through Socratic guidance embedded in-flow within their IDE. PC evaluates prompt quality across multiple dimensions and surfaces targe

In the October 2024 flash floods in Spain, social media posts falsely claimed authorities were concealing the number of casualties. The narrative centered on a flooded parking structure in Valencia, where hundreds of bodies were falsely claimed to be trapped. This narrative gained traction even after videos showing the premises had been evacuated. The post Narrative monitoring and argument-checking: Enhancing effectiveness in countering disinformation beyond fact-checking first appeared on HKS M

Not all misinformation consequences are equal. Faced by hundreds of thousands of false claims online and offline every day, fact checkers need a robust way to identify the important ones to check. This scalable model—used by fact checkers in trials in Europe, Africa, and the Middle East since 2024—helps forecast the potential imminent and cumulative harms of different false claims and is an early warning system for society that focuses efforts on factually false claims that cause real-world harm

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

Fact-checking has become a key response to disinformation during crises and conflicts, but its role is increasingly contested due to concerns about its effectiveness and its co-optation by different political actors. In polarized, high-choice environments, fact-checking is often embedded within partisan and state-aligned infrastructures, shaping validation and rejection of knowledge claims. The post Fact-checking in the multipolar AI order: Between epistemic sovereignty and ambivalence first app

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

Image guardrails are typically trained and evaluated under a fixed safety policy, implicitly treating safety as an intrinsic property of an image. Real deployments are different: the same image may be allowed in one product, restricted in another, and newly disallowed when a policy boundary changes. We study policy-adaptive image guardrailing, where a model must decide whether an image violates the currently supplied policy and generalize to held-out policy definitions. We introduce PolicyShiftB

Big Data & Society, Volume 13, Issue 3, July-September 2026. We offer a genealogy of recent artificial intelligence infrastructure investment, situating it within a longer-term strategy by Big Cloud firms to construct and dominate the cloud computing market. Based on an analysis of 12 years of financial data and ...

Contemporary institutions increasingly treat optimised procedures as decisions in their own right. This article advances an ontological limit claim for AI governance: moral judgment is constitutively personal and therefore non-delegable. Building on a minimal philosophical anthropology—person/thing distinction; irreducibility of phronēsis; the person as an end; responsibility as constitutive; and the capacity to initiate—we argue that algorithmic assistance can legitimately expand human delibera

Non-Player Characters (NPCs) require social cognition to enable intelligent and interactive behaviours within virtual environments. In gaming and other multi-agent systems, current NPC models often fall short in social intelligence and coordination because they cannot infer or anticipate the mental states of other agents. To address this gap, this paper introduces a novel social cognitive architecture that integrates Hierarchical Reinforcement Learning (HRL) with a Transformer-based Theory of Mi

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

In today’s world, where advanced technology such as artificial intelligence (AI) is reshaping human society, how can individuals’ autonomy in the use of such technology be understood? This paper focuses on the meanings of technological artifacts in individuals’ lives—that is, the subjective interpretation of their function within one’s plan for use—and attempts to formalize a set of human capacities to actively give meaning to technological artifacts as a form of personal autonomy. This study sh

High quality data is of vital importance for unlocking the full potential of AI for end users. Villalobos et al. stated in 2024 that finding new sources of such data is getting harder as most publicly-available human generated data will soon have been used. Additionally, publicly available data often is not representative of users of a particular system — for example, a research speech dataset of contractors interacting with an AI assistant will likely be more homogeneous, well articulated and s

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

Data visualizations can inform, explain, and sway public opinion and policy decisions. This course imparts design thinking and data ethics frameworks, along with practical software skills, to ...

Text-to-image diffusion models have achieved high visual fidelity and broad adoption, but remain vulnerable to safety violations when adversaries exploit them to synthesize illicit content. Existing alignment paradigms, from input sanitization to structural feature pruning, are largely organized around unsafe concepts explicitly exposed during filtering, editing, or localization. This leaves a blind spot for visual synonym attacks (VSA), a jailbreak where benign-looking prompts elicit prohibited

Faults on a cyber-physical system (CPS) are too rare and unrepresentative to characterise, or even to select a model on, so detection must instead model normal behaviour; the standard point-adjusted evaluation, however, rewards detectors that never do. CPS normal behaviour is the union of many imbalanced, curved, thin-fringed operating regimes rather than a single blob; we state this structure as ten assumptions (A1-A10), abbreviated Massive, Implicit, Imbalanced Multimodality (MIIM). We model t

The Nonprofit Leadership Certificate at Harvard Kennedy School will help you get to the next level as a leader in your organization. This certificate is designed for professionals working in nonprofit ...

Federated Learning (FL) enables multiple clients to collaboratively train machine learning models while retaining data locality, thereby enhancing user privacy. However, traditional FL frameworks rely on a centralized aggregation server and assume honest-but-curious clients, making them susceptible to both server-side inference and client-side poisoning attacks. Although recent work has explored secure and Byzantine-resilient FL protocols, they face a fundamental trade-off among privacy, integri

Recent advances in artificial intelligence (AI) and machine learning (ML) enable targeted optimization of emergency department (ED) operations. We examine how reworking an ED’s vertical processing ...