The World Health Organization has declared 2021–2030 the “Decade of Healthy Ageing”, aiming for the best quality of life through health as the population ages. Beyond healthy ageing, scientists are adopting artificial intelligence technologies for longevity science which can foreseeably enable humans to routinely live to 120 years and beyond. With such breakthroughs within reach, the challenges associated with longevity need to be considered, from the impact on the social system to the possibility of an international law right to longevity, along with associated considerations such as on sustainability. This article questions whether there already is, or should be, an international human right to facilitate considerably extended lifespans, along with other relevant legal frameworks.

Since the public release of ChatGPT in November 2022, the artificial intelligence (AI) landscape is undergoing a rapid transformation. Currently, the use of AI chatbots by consumers has largely been limited to image generation or question-answering language models. The next generation of AI systems, AI agents that can plan and execute complex tasks with only limited human involvement, will be capable of a much broader range of actions. In particular, consumers could soon be able to delegate purchasing decisions to AI agents acting as “Custobots.” Against this background, the Article explores whether EU consumer law, as it currently stands, is ready for the rise of the Custobot Economy. In doing so, the Article makes three contributions. First, it outlines how the advent of AI agents could change the existing e-commerce landscape. Second, it explains how AI agents challenge the premises of a human-centric consumer law which is based on the assumption that consumption decisions are made by humans. Third, the Article presents some initial considerations how a future consumer law could look like that works for both humans and machines.

An enduring access-to-justice crisis leaves most low- and middle-income people without meaningful assistance for civil legal problems. In response, several U.S. jurisdictions have experimented with licensing legal paraprofessionals—such as Limited License Legal Technicians (LLLTs)—to provide a circumscribed set of services directly to the public. Using Washington State’s pioneering LLLT program and its successors as a case study, this Article argues that paraprofessional reforms have under-delivered because they replicate key features of the traditional professional model: substantial educational prerequisites, supervised practice requirements, and high-stakes examinations that raise entry costs, limit supply, and constrain scalability.

The Article contends that modern AI changes the production function of routine legal work—particularly client intake, document preparation, and the translation of facts into legally relevant narratives—yet AI deployed directly to consumers poses serious risks, including error, bias, confidentiality threats, and jurisdictional mismatch, and it cannot reliably identify when a matter requires escalation to a lawyer. The Article therefore proposes an “AI–paraprofessional fusion” model: purpose-built, jurisdiction-specific AI tools paired with lightly trained human paraprofessionals who provide process guidance, verify and quality-control outputs, and triage cases for escalation when warranted.

Finally, because unauthorized-practice rules are state-created constraints that helped produce today’s scarcity, the Article argues that the AI infrastructure enabling this model should be developed and maintained as a public good—auditable, updateable, and broadly accessible—rather than left solely to private market incentives. This approach offers a scalable path for United States jurisdictions—and potentially others—to expand competent, lower-cost legal assistance while preserving safety through human oversight and clear escalation channels.

We stand at a curious moment in the history of law and technology. Nations around the world are scrambling to regulate or deregulate artificial intelligence, each convinced they are in a “race”—for dominance, for values, for the future itself. Brussels votes on comprehensive AI Acts. Beijing issues the world’s first copyright ruling on AI-generated content. Washington debates whether chatbots should have First Amendment rights. The underlying premise of this volume is that this framing as a zero-sum competition fundamentally misunderstands both the nature of AI and the task before us. The truth is more sobering and more hopeful: We are not racing against each other but experimenting together, trying to govern technologies that respect neither borders nor traditional legal categories. The real question is not who will “win” the AI race, but how we can learn from each other’s experiments fast enough to keep pace with systems that evolve by the microsecond. This Special Issue of the German Law Journal brings together fifteen contributions that demonstrate why comparative law has never been more essential—or more challenging. The authors span continents and legal traditions, from Beijing to Brussels, from Silicon Valley to Sydney.

This Article discusses China’s content moderation in the age of artificial intelligence. It first introduces two long-overlooked features of China’s content moderation: the medium-based model and the “No-Dispute” Policy. The former emphasizes that content moderation in China varies based on different media, while the latter argues that China’s content moderation is often content-neutral rather than being driven by ideology or having an official stance. The Article then summarizes the three main challenges artificial intelligence presents to content moderation: a shift in structure from the traditional “state v. citizen” dichotomy to the “platform–government–citizen” triangle; a transition in means from human review to algorithm-based and machine-based moderation; and stimulating a reimagination of traditional theories and doctrines of freedom of speech in terms of standards and classification. Finally, the Article takes online violence, one of the most prominent issues in contemporary Chinese content moderation, as a case study to examine specific issues in China’s content moderation in the era of artificial intelligence.

Systematic reviews (SRs) are critical for evidence-based research but are time-consuming and labor-intensive. The rapid expansion of academic publications further challenges the performance and applicability of existing screening and classification methods. While large language models (LLMs) present new opportunities for automation, limited research has examined whether they can achieve classification performance comparable to human reviewers in large-scale, multi-class settings. With the goal of improving classification performance, we proposed an LLM-based framework that leverages full-text key-insight extraction to enhance literature classification. We constructed a manually curated dataset of 900 articles from 17 published SRs to quantitatively evaluate the classification capabilities of LLMs. The results provided empirical evidence of LLMs’ potential in supporting large-scale SRs and introduced a practical pathway for improving efficiency and reliability in evidence synthesis. Empirical results showed that key-insight-based classification (KBC) significantly outperforms abstract-based classification (ABC). We implemented a confidence-weighted voting (CWV) mechanism using multiple LLMs to improve robustness. The CWV method achieved the highest macro F1-score of 0.796, substantially exceeding KBC (0.732), ABC (0.676), and unsupervised K-means clustering (0.446). By employing zero-shot LLMs, our approach demonstrated the potential for enhanced adaptability across diverse domains and classification tasks without requiring fine-tuning, demonstrating that a carefully designed pipeline can enable LLMs to achieve classification performance comparable to human reviewers.

Traditional perceptual models are ill-equipped for the high-dimensional data, such as text embeddings, central to modern psychology and AI. We introduce the double machine learning lens model, a framework that utilizes machine learning to handle such data. We applied this model to analyze how a modern AI and human perceivers judge social class from 9,513 aspirational essays written by 11-year-olds in 1969. A systematic comparison of 45 analytical approaches revealed that regularized linear models using dimensionality-reduced language embeddings significantly outperformed traditional dictionary-based methods and more complex non-linear models. Our top model accurately predicted human $(R^{2}_{CV} =0.61)$ and AI $(R^{2}_{CV} =0.56)$ social class perceptions, capturing over 85% of the total accuracy. These results suggest that “unmodeled knowledge” in perception may be an artifact of insufficient measurement tools rather than an unmeasurable intuitive process. We find that both AI and humans use many of the same textual cues (e.g., grammar, occupations, and cultural activities), only a subset of which are valid. Both appear to amplify subtle, real-world patterns into powerful, yet potentially discriminatory heuristics, where a small difference in actual social class creates a large difference in perception.

Arguably, recent and prospective developments within artificial intelligence are a fascination within contemporary technoculture. The dawning of a new era that is characterised by the various impacts of these technological and scientific advances leads to questions about the type of subject that will inherit and inhabit the consequences of these developments. This paper will examine the role that speculative fiction plays as a site of critical engagement in investigating some of the more urgent questions posed by the intersection between humans and technology, such as the social consequences of projected technologies and the possibilities of changing embodiment, and particularly how these issues prove to be of immense importance for the gendered subject. The essays contained within Jeanette Winderson’s non-fictional publication 12 Bytes: How We Got Here. Where We Might Go Next (2021) provide a perceptive insight into both the promises and the pitfalls of AI technology for the future female and embodied experience. Winterson’s thought-provoking contemplations will be read alongside her fictional novels, The Stone Gods (2007) and Frankissstein (2019), to consider how she utilises the genre of speculative fiction to explore existing representations of gender whilst working to define new transhuman subjects. A recurring theme throughout these novels is the way in which AI, despite its liberating and transcendent potential, is imagined as the inevitable perpetuation of female subjugation.

The rapid expansion of artificial intelligence has accelerated its adoption across organizational functions. However, existing reviews often adopt sectoral or technology-focused perspectives, limiting understanding of its implementation within core firm activities. This study addresses this gap through a systematic review of articles published in Web of Science and Scopus up to December 2025, following established methodological guidelines. A total of 160 peer-reviewed articles met the inclusion criteria. Findings reveal convergent patterns of adoption in human resources, marketing and customer services, logistics, and finance. Artificial intelligence enhances analytics, automates routine tasks, personalizes interactions, and supports decision-making. Human resources applications focus on recruitment and workforce planning; marketing relies on predictive analytics and conversational interfaces; logistics improves forecasting and supply chain resilience; finance strengthens risk assessment and process efficiency. The study proposes an integrative conceptual model and research propositions, highlighting cross-functional challenges in governance, organizational capabilities, socio-technical alignment, and responsible implementation.

Innovation in paediatric and adult congenital cardiology increasingly depends on collaboration among academia, industry, and professional communities. From this perspective, the author argues that clinical prediction represents a natural convergence point for these stakeholders, aligning safe, personalised care with economic incentives. The author discusses emerging evidence highlighting the promise of artificial intelligence-driven prediction across various cardiovascular domains, while highlighting current limitations related to narrow scope, static design, and weak integration into clinical decision-making. Medicine-based evidence and a high-quality, inclusive data infrastructure may help address these gaps. Together, these approaches, along with stakeholders upholding their responsibilities, define a path towards predictive innovation.

This study investigates employees’ perceptions of artificial intelligence (AI) in the workplace, using data from 1,224 working adults across two samples. Drawing from an extended version of the Technology Acceptance Model, we examine how employees’ trust in AI and their perceptions of AI’s usefulness and ease-of-use at work shape their affective attitudes toward using AI, which in turn influence their intentions to adopt AI in their job. Perceived usefulness and trust in AI predicted employees’ intentions to adopt it at work via affective attitudes toward using AI. The findings for perceived ease-of-use were inconsistent, suggesting potential workplace-specific implications of this pathway. None of the relationships differed by gender, education, or leadership status. The findings bridge the technology adoption and organizational science literature to offer theoretical insights, practical implications, and future research directions for facilitating employees’ intentions to adopt AI at work.

The global demand for artificial intelligence (AI) is fuelling a rapid expansion of data infrastructure, an industry that is notoriously water-intensive. This growth creates a critical, yet understudied, nexus between digital expansion and hydrological systems, particularly in ecologically vulnerable regions. This study applies a spatially explicit framework to quantify the water footprint of AI data centres in Brazil, a nation heavily reliant on drought-sensitive hydropower. Our method integrates datasets on data centre locations, regional hydrological cycles, power generation sources and watershed-level water stress indices to model both direct (cooling) and indirect (energy generation) water consumption. Our key finding is that the AI infrastructure cluster in the São Paulo metropolitan region, with an operational IT load of ~550 MW, has an estimated annual water footprint of 16.1 million cubic metres. A significant portion of this, over 46%, is indirect “virtual water” consumed through hydropower generation, establishing a direct feedback loop where data centre demand stresses water and energy systems already compromised by climate change. This article concludes that the environmental cost of AI extends beyond carbon to include water, a cost disproportionately borne by biodiverse regions. We call for a paradigm shift in tech policy and corporate sustainability to include metrics of water neutrality and watershed resilience, in alignment with global sustainability goals.

This article examines the transformative impact of large language models (LLMs) on online content moderation, revealing a critical gap between platforms’ rule-based policies and their AI-driven enforcement mechanisms. Using Facebook’s hate speech moderation policies and practices as a case study, we identify a paradox: while content policies are increasingly rule-oriented, AI-driven enforcement seems to operate in a standard-like manner. This disconnect creates transparency, consistency and accountability challenges relating to the delineation of online freedom of expression that are not addressed in the literature, and require attention and mitigation. In this specific context, we introduce the concept of ‘rules by the millions’ to describe how AI systems actually operate through generating vast networks of micro-rules that evade traditional regulatory oversight. This phenomenon disrupts the conventional rules-versus-standards framework used in legal theory, raising urgent questions about the adequacy of current AI governance mechanisms. Indeed, the rapid adoption of LLMs in content moderation has outpaced the human capacity to monitor them, creating a pressing need for adaptive frameworks capable of managing the evolving capacities of AI.

This study investigates the use of large language models (LLMs) to classify question utterances within verbal design protocols according to Eris’ (2004) taxonomy. We evaluate the performance of two proprietary LLMs – OpenAI’s GPT-4.1 and Anthropic’s Claude Sonnet 4.5 – across experiments designed to assess classification accuracy, sensitivity to prompt configuration and in-context learning (ICL), and generalization across datasets and models. Using two human-coded datasets of differing size and quality, we measure alignment between LLM-generated labels and human judgments at both question category and subcategory levels. Results show that both LLMs achieved moderate to strong alignment rates at the category level (up to 85.7% for GPT-4.1 and 82.9% for Claude Sonnet 4.5), with substantially lower alignment at the more granular subcategory level. Performance differences across prompt configurations and ICL conditions were small, indicating robust generalization across datasets and transferability of prompt designs. While these results suggest that LLMs can effectively support scalable question classification, human judgment and oversight remain essential. Future research should explore the development and evaluation of alternative hybrid human–LLM workflows in protocol analysis, as well as the use of smaller or open-source models to address data privacy concerns.

When foundation models analyze political content, do they use demographic characteristics as shortcuts for ideological attribution? We conducted detailed experiments with GPT-4o-mini and validated key findings across GPT-4o and LLaVA, using identical, ideologically neutral campaign advertisements with systematically varied candidate demographics. All models consistently attributed more liberal ideologies to women than men. These effects exceeded real-world gender differences from a nationally representative survey. However, racial associations differed by model: strong in GPT-4o-mini (where Black candidates received substantially more liberal attributions), attenuated in GPT-4o, and insignificant in LLaVA. These demographic effects persisted across temperature settings, prompt variations, and even explicit debiasing instructions in GPT-4o-mini. Our findings reveal that visual demographic features can shape AI outputs in ways that vary across models, with implications for applications such as content classification.