Insights

The first informal exchanges on AI in the Military Domain (AIMD) took place at the United Nations in Geneva in June 2026, marking progress after years of diplomatic efforts on military AI and lethal autonomous weapons systems (LAWS). These discussions indicate the potential to establish a permanent, consolidated forum to coordinate global governance and international norms on military AI. However, achieving unified multilateral cooperation remains challenging due to fragmented existing initiatives and major powers' reluctance to engage in collective normative frameworks.
AI in the Military Domain: Multilateralism at a Crossroad
US economic growth is exceeding 2%, yet consumer and business confidence surveys reflect recession-level pessimism, revealing a significant disconnect. Key surveys like those from the Conference Board, University of Michigan, and ISM have frequently signaled downturns that do not materialize. This discrepancy arises from falling survey response rates, social media-driven discontent, political polarization, and inherent survey biases. Notably, rising inequality means surveys—weighted equally across respondents—fail to capture growth driven primarily by the wealthiest consumers, who now dominate half of all spending. Hence, economic surveys have lost relevance as indicators of actual economic conditions.
Why economic surveys have lost their relevance
A foresight study by Adam Zable and Stefaan Verhulst convened experts across multiple fields to identify seven key trends in data governance amid the rise of AI: strain on the open-data model; machine-centric, AI-mediated data ecosystems; inference transforming governance foundations; unsustainable data infrastructure; fragmented governance across jurisdictions; growing emphasis on sovereignty and strategic control; and need for better incentives in data sharing. These interconnected signals highlight how data governance increasingly intertwines with AI governance, digital infrastructure, economic strategy, democratic resilience, and geopolitical dynamics, emphasizing the need for anticipatory governance approaches to address emerging risks and opportunities ahead of lock-in effects.
ICT, Health
Anticipatory Data Governance in the Age of AI: Emerging Signals in Data Access, Reuse, and Sovereignty
Much of England is now in drought, threatening water supplies and severely impacting crop conditions, with farmers reporting dry soils and poor harvests. The UK food system is increasingly vulnerable due to aging infrastructure, heavy reliance on imports—especially for fruit and vegetables—and limited investment in renewal. Despite stable overall food self-sufficiency, only about 16% of fruit and slightly over half of vegetables are domestically produced. To better manage food supply risks, a new UK food resilience index is proposed, combining diverse indicators such as production capacity, import dependency, infrastructure, and investment. This would enable early detection of vulnerabilities and enhance the system’s ability to absorb shocks like droughts and geopolitical disruptions.
Africa | Financial services, Health
UK drought: crop harvests at severe risk of collapse
Millions of Americans face mental illness annually, but therapy remains costly and clinical shortages cause long waits. AI-powered mental health tools, including therapeutic chatbots and apps, are expanding to fill this gap, offering increased access, lower cost, and social support. However, these tools carry significant risks without clear regulation or standardized evaluation, including delivering substandard care and failing to identify crises. Legislative efforts are mostly state-driven with limited federal focus, complicated by rapidly evolving AI models and a lack of rigorous, validated research on their effectiveness and safety.
The Complexities of Governing Mental Health AI
Open data has expanded democratic access to information and enabled greater scrutiny and reuse, but governance challenges now extend beyond mere access. Data gain power through processes after publication—such as cleaning, linking, and modeling—that inform decisions often made by external entities. This "closed inference" creates an asymmetry where data are open, but the capacity to interpret, validate, and contest derived conclusions is concentrated and less accountable. As a result, public institutions may lack control over the authoritative meanings assigned to data, raising concerns about transparency and governance in decision-making based on open data.
ICT, Health
Open Data, Closed Inference
Digital public goods (DPGs) and the open knowledge commons face significant challenges from large-scale, automated extraction by AI crawlers operating without reciprocity, causing rising infrastructure costs and undermining trust. To address this, new resources have been developed collaboratively by DPG product owners, including a position paper with recommendations for engaging commercial AI users to respect licenses and contribute back, and a playbook outlining technical and policy mechanisms to protect open resources. The playbook introduces a six-layer defense framework combining technical, legal, regulatory, and institutional measures, emphasizing that safeguarding against AI exploitation requires coordinated ecosystem-level solutions beyond code-based protections.
ICT
Navigating the Paradox of Open: New Resources for Defending Data and Content DPGs in the Age of AI
The largest technology firms are evolving into corporate sovereigns by dominating critical AI infrastructure, including chips, data centers, and frontier AI model development. This concentration creates vertically integrated, mutually dependent entities with extensive influence and reduced external oversight. AI companies control their own regulation processes and can evade jurisdictional accountability through complex digital and corporate strategies. In global health, reliance on AI solutions from these firms risks creating dependencies on supply chains outside local control, raising significant concerns about governance, equity, and sustainability.
ICT, Health
Global health suffers when corporate AI sovereigns reign
AI industry frequently uses the prefix "co" (as in co-pilot, co-worker, co-creator) to suggest AI systems work collaboratively with humans, implying shared control and joint agency. This linguistic choice creates strategic ambiguity, allowing products to be marketed as both autonomous to investors and subordinate to users, reassuring human oversight. A notable example is Microsoft's Copilot, which positions AI as an assisting partner rather than a leader, fostering dependency while maintaining a perception of human control. However, concerns arise as powerful AI could lead humans to disengage, effectively ceding control to AI systems. This framing contributes to increasing anthropomorphism of AI, influencing how people perceive these tools and signaling a shift toward co-existence with AI beyond human dominance.
Why the AI industry is so fond of the prefix ‘co’ – by an expert in linguistics
Organizational challenges often stem from difficulties promoting open upward communication, as employees may choose silence due to fear, fatigue, or belief that speaking up is futile. Research in higher education shows that even small perceived career costs for dissent can filter out employees who challenge norms, impacting leadership pipelines by favoring those who avoid conflict and prioritize compliance. Incentive structures that emphasize alignment and penalize dissent may suppress innovation and resilience by discouraging constructive challenge. Organizational culture and HR practices play key roles in shaping whether employees voice concerns and who advances into leadership roles.
What happens when employees who speak up are held back?
Human development inequality in Asia remains significant, with Human Development Index losses due to inequality ranging from under 10% to over 35%. Education and income disparities are the main contributors, especially pronounced in South Asia compared to East Asia and the Pacific. Gender equality plays a crucial role in reducing these losses through women's education, workforce participation, and empowerment. Strong government effectiveness correlates with lower inequality levels, emphasizing the importance of institutional quality. Addressing educational and income inequalities, promoting gender-inclusive development, and improving governance are essential policy priorities.
Africa, ASEAN, Latin America | Health
Explaining Human Development Inequality in Asia: Evidence and Policy
AI generates predictions that humans incorporate alongside their own analyses and other data sources to make decisions. The economic value of AI depends on this interaction within the broader decision-making environment. Optimal AI training focuses not simply on maximizing prediction accuracy but on how AI predictions function as part of a "composite experiment" with human users. This approach considers factors like economic variables, user heterogeneity, and the conditions under which AI predictions are applied, demonstrating that training strategies must account for the decision context to maximize value.
Education
Training AI for When Humans Will Use It
Media freedom and journalism face global threats including AI-driven disinformation, rising censorship, violence against journalists, and declining funding. Independent, ethical journalism plays a crucial role beyond democracy, positively impacting economies by reducing corruption and fostering growth, enhancing national security through resilience against disinformation, and improving crisis response by providing life-saving information. Research shows journalism is highly cost-effective, yielding significant public savings and acting as a vital preventive element in maintaining a healthy information infrastructure.
Africa | ICT, Health
The value of journalism: global evidence on why media matters to economies, national security and crises
Conflict zones facilitate the spread of infectious diseases by disrupting healthcare, vaccination, sanitation, and displacing populations, making outbreaks more frequent and harder to detect. Analysis of WHO disease outbreak reports from 1996 to 2024 shows increased reported outbreaks of polio and cholera during armed conflicts, especially in sub-Saharan Africa, linked to damaged infrastructure and reduced immunization. Conversely, influenza outbreaks are reported less during conflicts, likely due to damaged surveillance and reporting systems rather than reduced transmission. Overall, conflicts increase disease risks while impeding timely detection and response.
Africa | Health
Conflicts can spread infectious diseases while making outbreaks harder to detect
For two decades, the open data community has adhered to the FAIR principles—Findable, Accessible, Interoperable, and Reusable—but these are insufficient for the AI era. A fifth principle, Ready for AI, is needed. Corporate discussions on AI data readiness, such as McKinsey’s insights on scaling AI impact, offer valuable lessons for public data stewards. While companies focus on reliability and cost, public-interest data stewardship must also emphasize equity, openness, and accountability. Thus, some corporate practices can inform the extension of FAIR principles, but adaptations are necessary to align with public sector values.
What Open Data Stewards Can Learn from Corporate Guidance on AI-Ready Data
Andy Burnham’s rapid rise from Metro Mayor of Greater Manchester to UK Prime Minister follows a familiar European pattern where city leadership serves as a stepping stone to national power. Burnham leveraged devolution, infrastructure investment, and public service control to build a strong platform. European examples show three common pathways from mayor to national leader: the "boomerang" from national to local and back, the "leapfrog" from city leadership without prior national roles, and the "linear ascent" through gradual progression. Across Europe, city leadership holds significant prestige and power, with many politicians moving fluidly between local and national offices. However, local popularity does not always guarantee national success.
Africa
A bee line to Downing Street – Andy Burnham and the power of the European mayor
The European Commission fined Google €890 million for violating the EU Digital Markets Act, underscoring Europe's strict regulation of US big tech. Despite being one of the largest fines under the DMA, the penalty amounts to less than 0.3% of Google's annual revenue, raising doubts about its impact on company behavior. Previous EU fines on Microsoft and Google show a pattern: fines lead to technical or contractual changes, but disputes over effective compliance persist. The key challenge is ensuring fines translate into genuine market changes, such as increased competition and consumer choice. EU regulators face criticism for slow proceedings and are exploring reforms to improve enforcement speed and transparency amid rapid technological advances.
The EU has fined Google a billion dollars for ignoring competition rules – but will it have any effect?
When you hit “delete,” data is often not fully erased but merely marked as reusable space, making recovery possible. True deletion requires methods like overwriting, physical destruction, or encryption key destruction, which can be costly or impractical, especially with cloud storage. Deleting apps does not remove associated accounts, leading to "zombie accounts" vulnerable to breaches. AI systems complicate deletion further, as trained data is hard to "unlearn" despite legal rights like the EU’s GDPR "right to be forgotten." The proposed concept of “explainable deletion” aims to make deletion processes transparent, giving users clear, manageable information and control, potentially increasing trust and compliance.
What really happens to your data when you click ‘delete’?
Cross-border e-commerce is rapidly transforming trade in Asia and the Pacific, where the region accounts for two-thirds of global e-commerce sales. However, significant disparities remain, with one-third of the population lacking connectivity and many MSMEs unable to fully engage in digital trade. Bridging the digital divide, enhancing regional cooperation, harmonizing standards, and adopting holistic strategies—such as regulatory reforms, trade facilitation, and investment in digital infrastructure and skills—are essential to support MSMEs and promote inclusive, broad-based economic growth.
Africa, ASEAN, Latin America | Health, ICT, Financial services
Advancing Digital Trade and Cross-Border E-Commerce in Asia and the Pacific
Defining openness for foundation models (FMs) is challenging, as traditional open source software definitions do not readily apply to AI systems. Establishing shared nomenclature across researchers, developers, policymakers, and civil society is crucial for building common understandings, norms, benchmarks, and best practices. This approach highlights the variety of design choices affecting openness throughout the AI stack, beyond just model weights. A proposed descriptive framework evaluates how different components contribute to openness, supporting normative definitions without prescribing fixed requirements. Insights build on a 2024 workshop with experts from diverse AI sectors, aiming to clarify what openness should mean for foundation models today.
Unpacking Open Source Artificial Intelligence: Toward a Framework for Openness in Foundation Models

Forum Spaces

© 2026 All rights reserved
Privacy Policy & Terms of Service