Using a documentary biography informed by Biographies of Artefacts and Practices (BOAP), this article traces how branches of GPT-4 and DeepSeek-R1 became reachable through regulatory pathways in the European Union, United States and China. Drawing on 132 documentary observations, it reconstructs five cross-site configuration moments from 2022 to 2025, supplemented by a bounded mid-2026 status check. The matched comparison covers hosted and API branches of both releases, while DeepSeek-R1's repository, derivatives and public serving routes extend the analysis to redistributed open-weight branches. Governance initially operated through inherited platform, algorithmic-service, data and compute controls before more explicit provider-, model- and output-oriented pathways emerged. Provider-controlled services preserved identifiable actors and control points, whereas open-weight redistribution allowed model lineage to persist across repositories, modifiers, hosts, deployers and platforms beyond the original developer's continuing control. Organisational responses further reconfigured later control points. The article develops generative AI in context as a BOAP-informed extension for model lineages whose technical continuity persists while distribution relationships and organisational control diverge. Branching regulatory biography captures this divergence, while regulatory reachability identifies which instruments can address which actors through which control points for a particular branch and time. Distributed-model governance therefore requires differentiated responsibilities and portable records, while treating documentation as support for, rather than a substitute for, accountable legal actors.
Despite longer lifespans, many people keep feeling “time poor,” often struggling to complete daily tasks. This challenge is most pronounced among women, who frequently have disproportionate domestic labor responsibilities, and low-income individuals, who frequently lack the resources to outsource time-intensive activities. As time poverty reinforces and perpetuates social inequalities, policies to alleviate it, such as reducing time spent on errands like queuing for goods and services, can play a key role in addressing these disparities. In this context, we explore the potential of expanding broadband adoption as a possible policy solution as it may enable more efficient access to goods and services online, thereby reducing waiting times. Integrating broadband data with diary records from the American Time Use Survey, we construct a 14-year panel dataset (2008–2021) and estimate a fixed effect model. We further examine the impact for broadband by gender and income. Findings contribute to policies intending to alleviate time poverty by demonstrating that living in connected counties reduces time spent waiting for goods and services, with the effect being particularly pronounced among women. However, we also highlight the need for complementary policy measures to fully address persistent disparities. Our results are robust to different model specifications and an instrumental variable approach.
Mobile money (MM), as a critical digital financial infrastructure, has become increasingly central to sustainable development in the Global South. Although its socioeconomic benefits have been extensively documented, the role of MM in advancing universal energy access has been underexplored, especially in sub-Saharan Africa, where electrification rates are persistently low. Within this regional context, Kenya represents a compelling case of the digital-green-policy nexus: the widespread diffusion of MM converges with off-grid solar (OGS), a green energy solution, under the Kenya Off-Grid Solar Access Project (KOSAP), a policy initiative aimed at expanding household electrification in underserved counties. This nexus offers a concrete setting to examine the twin green-digital transition across different policy environments. Using nationally representative data from the Kenya FinAccess Survey 2024, this study investigates whether and how MM use facilitates household OGS adoption in KOSAP and non-KOSAP counties. We find that MM use significantly increases the likelihood of OGS adoption, with no significant difference across the two policy environments. Mediation analysis reveals policy-dependent digital financial pathways: mobile savings (loans) appear more salient in KOSAP (non-KOSAP) counties. These findings provide micro-level evidence of digital sustainability and the twin transition in the Global South and underscore the need for an integrative policy framework that coordinates policymaking across the digital finance and energy sectors.
As artificial intelligence (AI) assumes growing strategic significance worldwide, attention has increasingly shifted from discrete technologies to the ecosystems within which AI is developed, deployed, and governed. While existing scholarship has devoted considerable attention to standard-setting battles and export-control regimes, the ecosystem-level dynamics of AI competition remain under-theorised. Drawing on a National Innovation Systems (NIS) framework enriched with a novel "openness–security tension" analytical dimension, this study examines how Chinese AI policy simultaneously pursues ecosystem openness and security control, and how this dual imperative shapes the international trajectory of China's AI ecosystem. Through a mixed qualitative methodology combining policy content analysis of 63 central-government documents (2017–2025), NIS structural mapping, and nested case studies of three AI ecosystem layers—open-source AI, AI platform services, and AI data governance—the study identifies a pattern of "selective internationalisation." Findings reveal that the three ecosystem layers occupy distinct positions along an openness–security spectrum: open-source AI exhibits the highest degree of international engagement, AI platform services display a mixed pattern constrained by data-compliance barriers, and AI data governance demonstrates the strongest inward orientation. These divergent trajectories tend to vary inversely with each layer's degree of ideological sensitivity—the degree to which a layer's operations intersect with state concerns about political content and information security. The study further suggests that China's tendency toward AI "internal circulation" can be interpreted as less a deliberate strategic choice than an unintended cumulative consequence of security controls propagating through mutually reinforcing NIS elements, a dynamic illuminated through engagement with Heilmann's concept of unorthodox policymaking. These findings contribute to the literature by extending NIS theory to incorporate security considerations and institutional tensions, shifting the analytical focus from standard wars to ecosystem competition, and challenging binary narratives of Chinese AI governance.
As generative artificial intelligence (GenAI) becomes increasingly embedded in digital communication services, telecommunications operators and other GenAI providers face growing copyright governance challenges arising from AI-assisted content creation. These include determining the copyrightability of AI-generated outputs, establishing standards for assessing copyright infringement, and allocating copyright infringement liability between users and GenAI providers. While existing scholarship has largely treated these issues separately, this article develops an integrated analytical framework by analysing recent Chinese judicial decisions concerning GenAI-assisted works. We demonstrate that copyright law can adapt to these challenges when courts identify a demonstrable ‘mapping relationship’ between user prompts and generated outputs. Under this approach, GenAI-assisted outputs may be eligible for copyright protection if users provide sufficient evidence of such a mapping relationship to establish human authorship; copyright infringement may occur when subsequent users replicate prior prompts to generate substantially similar outputs; and direct liability rests with the user rather than the GenAI provider. Moreover, these findings have broader governance implications than copyright doctrine. For telecommunications companies integrating GenAI into customer-facing services, this analysis clarifies how user agreements, evidence-preservation mechanisms, and user logs may facilitate copyright governance and risk allocation. For legislators, courts, copyright authorities, and digital economy regulators, this proposed framework supports the development of a coherent rights-and-liability regime that balances innovation, legal certainty, and accountability in GenAI-enabled digital content ecosystems.
As Internet traffic increases rapidly, network operators face the challenge of simultaneously expanding coverage, deploying next-generation networks, and increasing network capacity to accommodate continued traffic growth. Such factors have triggered a policy debate known as the “fair share debate” between content and application providers (CAPs) and Internet service providers (ISPs), potentially affecting cost conditions that change incentives for deploying broadband in rural areas, which are characterized by a high-cost, low-revenue structure. We quantitatively analyze the impact of traffic growth on network investment cost structures and cost conditions that shape incentives for rural broadband deployment under surging traffic. We find that incumbent operators may perceive network capacity investment as inefficient once traffic exceeds 3.4 times the current level. When traffic ranges between 110% and 415% of the current level, some rural areas may experience marginal traffic costs as network capacity investment falls below the average cost of coverage deployment per building, potentially weakening cost-related incentives for rural broadband investment. Subsidizing broadband deployment can adjust this range, suggesting that such support remains a valid policy instrument, even under conditions of explosive traffic growth. Policymakers should consider an integrated approach to network policy design to enable network operators to sustain rural broadband investment in the face of persistent CAP–ISP conflicts. This study links traffic growth, network investment costs, and rural broadband expansion, providing empirical evidence regarding network operators’ investment decisions and policy insights relevant to cost-sharing discussions among traffic-generating entities and the design of rural broadband expansion policies.
This study examines publication outcomes for research presented at the Telecommunications Policy Research Conference (TPRC) using conference and publication data from 1994 to 2022. Focusing on academic publication as one observable and structured stage of scholarly record entry, we examine whether publication rates and time to publication have shifted over time; whether factors such as gender, disciplinary affiliation, or community roles are associated with publication outcomes; and the extent to which research presented at this subfield-specific conference appears in a broader range of publication outlets.Approximately one-third of presented papers are ultimately published in academic journals, and publication rates remain relatively stable across the period studied. Average time to publication is descriptively lower in most recent years, falling from roughly 1.85 years in 2000 to 0.95 years in 2022, though formal regression results do not indicate a robust linear time trend. Publication timing and outcomes differ across disciplines, with economics and communication journals exhibiting longer publication timelines than policy journals, while law journals publish more quickly. After accounting for disciplinary differences, journal tier is not a significant predictor of publication timing. Women are underrepresented in formal conference roles and are less likely to publish in peer-reviewed journals. Formal membership in TPRC governance or editorial networks shows no detectable association with publication likelihood.By treating academic publication as a measurable but incomplete indicator of scholarly entry into the formal scholarly record, this study provides a systematic account of how policy-oriented, interdisciplinary research presented at TPRC appears in academic journals. Broader forms of policy engagement, such as regulatory filings, advisory work, and participation in policy processes, remain outside the scope of this analysis and are difficult to observe directly.
Satellite internet has become a cross-domain system with both strategic and infrastructure attributes. Existing studies have examined its technical trajectories and geopolitical implications, but provide less systematic evidence on how innovation-policy mixes evolve across instruments, time and space. Focusing on China, this paper constructs a corpus of 132 central and subnational policy documents and codes 363 substantive policy-instrument reference points using a satellite-internet adaptation of the Rothwell and Zegveld supply-, environment- and demand-side framework. It combines manual coding, co-word network analysis and spatial comparison to map stated policy attention. The results show a persistent supply-side dominance, with attention concentrated on infrastructure, operation services and application-oriented industrial development, while regulation, finance and ecosystem coordination remain less visible. Spatially, policy supply is concentrated in major industrial regions, but lower-centrality provinces often display locational or industrial specialisation rather than simple imitation. The findings offer a replicable text-based approach to analysing policy mix and attention in cross-domain infrastructure systems and identify the coordination mechanisms needed for the next stage of satellite-internet development.
The network industries are increasingly entered by conglomerate subsidiaries whose parent companies cross-subsidize below-cost pricing from revenues in unrelated industries. This paper introduces conglomerate leveraged entry (CLE) — a formally defined entry type that renders limit pricing, capacity commitment, and tacit collusion simultaneously ineffective as deterrence mechanisms. We develop a three-phase equilibrium framework establishing the CLE welfare paradox: consumer surplus is highest during the price war, lower under pre-entry oligopoly, and lowest post-reconcentration. Using 418 circle-quarter observations from India's telecommunications market (Q1 2014–Q1 2025), we exploit Reliance Jio's staggered 4G rollout as a quasi-natural experiment. A Callaway-Sant'Anna estimator yields ATT = 171.9 HHI points (p = 0.034). Circle-level HHI rose 68.3 percent above its pre-entry mean. ARPU collapsed 42.8 percent before recovering 19.5 percent above the pre-entry peak validating the welfare paradox empirically. Treatment effects are heterogeneous: competitive in large urban circles, strongly reconcentrating in rural ones with direct implications for universal service regulation. The findings generalise to any capital-intensive network industry subject to conglomerate-backed entry.
