Who Owns an AI Enabled Future?

Introduction

An AI-enabled human bloom would not be decided by model capability alone. It would be decided by power: who controls the frontier systems, who can afford the compute, who sets the rules, who bears the risks, and who receives the gains. Advanced AI could accelerate medicine, education, clean energy, public services and scientific discovery, but those benefits will not automatically spread through society. Without deliberate governance, the same technology could deepen dependence on a small number of firms, strengthen military and surveillance capacity, widen the gap between rich and poor countries, and turn public institutions into customers of systems they do not properly understand or control.

The central governance question is therefore not “AI or no AI?” but “under what institutions?”. A blooming future would need more than safety testing and innovation policy. It would need competition rules, public compute, transparent public-sector use, worker voice, democratic oversight, international coordination, and credible ways to share value without handing dangerous capabilities to reckless actors.

Why ownership matters for AI abundance

AI abundance is often described as a world where intelligence becomes cheap and widely available. That phrase can hide a hard fact: advanced AI is not only software. It depends on scarce inputs, including high-end chips, data centres, cloud platforms, electricity, specialist talent, proprietary data, distribution channels and regulatory access. If those inputs remain concentrated, “abundant intelligence” may arrive as a subscription service controlled by a few powerful organisations rather than as a broad civic resource.

This is already visible in the structure of the AI economy. Foundation models require enormous investment in compute, data and skilled labour. Research on the economics of frontier model training finds that the cost of the largest training runs has been rising rapidly, with AI accelerator chips and staff costs among the biggest expense categories; if current trends continue, the largest training runs could cost more than a billion dollars by 2027. That does not prove that only a few firms can ever build useful AI, but it does mean that frontier capability is pulled towards actors with exceptional access to capital and infrastructure.[arXiv]arxiv.orgarXiv The rising costs of training frontier AI modelsThe rising costs of training frontier AI modelsMay 31, 2024…Published: May 31, 2024

Competition authorities have begun to treat this as a structural issue rather than a normal software market. The UK Competition and Markets Authority has investigated public cloud infrastructure, and OECD analysis of cloud markets notes that Amazon and Microsoft together have been reported as holding up to 80% market share in some large OECD economies, with Google commonly the third-largest provider. Cloud concentration matters for AI because many firms, universities and public agencies cannot train or deploy advanced systems without renting infrastructure from the same small group of providers.[GOV.UK]GOV.UKCloud services market investigationOctober 5, 2023 — Cloud services market investigation. The Competition and Markets Authority (CMA) investigated the supply of public clou…Published: October 5, 2023

The deeper concern is vertical power. A company that controls cloud infrastructure may also invest in frontier model developers, sell AI services, host enterprise data, offer workplace software, run app ecosystems and influence standards. That can make AI diffusion fast, but it can also make whole sectors dependent on bundled systems, private pricing decisions and opaque contractual terms. Economic Policy research on foundation models argues that the most capable models may show tendencies towards concentration, and that competition policy should prevent power at the model layer from propagating into downstream markets while also requiring standards on safety, privacy, reliability and interoperability.[arXiv]arxiv.orgarXiv The rising costs of training frontier AI modelsThe rising costs of training frontier AI modelsMay 31, 2024…Published: May 31, 2024

For the AI bloom thesis, this is not an anti-business point. Private firms may be essential to building frontier systems, scaling products and financing infrastructure. The issue is whether societies let the resulting platform power become the default constitution of the AI age. If a small set of companies becomes the gatekeeper for education tools, scientific assistants, medical discovery platforms, robotics systems and public administration, then “who owns the future?” becomes a practical question about prices, access, accountability and democratic control.

Commercial concentration: the promise and the trap

Commercial concentration has an ambiguous role in an AI-enabled future. On one side, large firms can marshal capital, talent and engineering discipline on a scale that universities, charities and most governments cannot match. They can build data centres, harden systems against attacks, run large-scale evaluations, maintain user-facing products, and integrate AI into tools people already use. A fragmented ecosystem of underfunded actors would not necessarily be safer or fairer.

On the other side, concentration can shape the direction of progress. Companies optimise for revenue, strategic advantage and shareholder value, not necessarily for the public’s highest-value needs. A society may need AI for neglected diseases, disability access, climate adaptation, local-language education, public-interest science and cheaper legal help. The most profitable deployments may instead be advertising, finance, surveillance, enterprise automation, entertainment, coding tools and military contracting.

This divergence matters because AI’s largest gains may come from general-purpose infrastructure that many sectors build upon. If access is expensive, locked down or skewed towards high-margin customers, the long-run benefits can be delayed or channelled towards already powerful users. The risks are not limited to monopoly pricing. They include vendor lock-in, dependence on proprietary systems, loss of public technical capacity, closed safety evidence, restrictive licensing, and a world where universities and smaller firms become downstream users rather than independent contributors.

Several policy tools aim to reduce this dependence without pretending that government can simply replace the frontier labs. Public compute is one of the clearest examples. The US National Artificial Intelligence Research Resource pilot connects researchers and educators with compute, data, models, software and expertise, with goals including accelerating AI-powered discovery, expanding the AI workforce and advancing interpretability, security and trust.[NSF]nsf.govSource details in endnotes. - U.S. National Science Foundation The UK’s AI Research Resource similarly provides AI-specialised compute capacity through advanced supercomputers for researchers, academia and industry.[GOV.UK]GOV.UKA I Research ResourceA I Research Resource

Public compute is not a magic equaliser. The largest private clusters may still exceed public resources, and access programmes can be slow, bureaucratic or skewed towards already well-connected institutions. But they change the bargaining position of society. They let public-interest researchers test models, build open tools, study safety, work on local needs and train talent without relying entirely on the commercial cloud. The Ada Lovelace Institute describes “public compute” as government-funded access to compute through hardware, vouchers, cloud credits or public supercomputing, and stresses that the meaning of “public” is contested: it can mean publicly funded, openly accessible, directed towards public-interest research, or some mixture of these.[Ada Lovelace Institute]adalovelaceinstitute.orgcomputing commonscomputing commons

A practical governance agenda for commercial concentration would not try to freeze the industry in place. It would ask sharper questions: Are cloud and model markets contestable? Can customers move data and workloads? Are public agencies buying systems they can audit and exit? Are safety evaluations independently reproducible? Are open models available where they are safe and useful? Are procurement rules rewarding interoperability rather than dependency? These are the plumbing questions behind a future that looks abundant rather than rented.

Military and political power cannot be an afterthought

The same features that make advanced AI useful for science and public services also make it strategically valuable to states. AI can help analyse intelligence, automate cyber operations, improve logistics, guide drones, accelerate weapons design, generate propaganda, and compress decision-making in crises. That is why governance of AI power cannot be left to consumer protection or workplace policy alone.

Military AI raises a particularly sharp version of the bloom dilemma. Civilisation may benefit from AI systems that help defend against biological threats, cyberattacks, nuclear escalation, disinformation or hostile uses of autonomous systems. But the pursuit of military advantage can also create arms-race dynamics: faster deployment, less transparency, weaker public debate, and pressure on companies to relax safeguards. European Parliament research on defence and AI notes the absence of a unified international framework, contrasting flexible US approaches with more human-centric European regulatory traditions, while warning about accountability, international humanitarian law and reduced human oversight.[European Parliament]europarl.europa.euEuropean Parliament Defence and artificial intelligenceEuropean Parliament Defence and artificial intelligence

Current international efforts are real but limited. The US-led Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy encourages states to support responsible military AI practices.[U.S. Department of State]2021-2025.state.govSource details in endnotes. SIPRI’s 2025 review describes the governance of autonomous weapon systems as centred for years on discussions under the Convention on Certain Conventional Weapons, alongside newer forums for broader military AI.[SIPRI]sipri.org12. Artificial intelligence and international peace and security12. Artificial intelligence and international peace and security These initiatives create norms, but they do not yet amount to a strong global regime that can reliably prevent unsafe deployment or escalation.

Frontier AI also sits inside wider geopolitical competition. Export controls on advanced chips, national AI strategies, security testing, data-centre location decisions and government partnerships with AI labs all shape who can build powerful systems. Compute governance is increasingly treated as a lever because advanced chips and large training clusters are easier to monitor than abstract “algorithms”. Lawfare analysis by Lennart Heim argues that compute governance can increase visibility into AI progress, shape allocation of key inputs and enforce rules around development and deployment, while also warning that such measures are technically and politically difficult to design well.[Default]Lawfareto govern ai we must govern computeto govern ai we must govern compute

The danger is that security policy becomes the whole of AI governance. If governments treat AI mainly as a race for dominance, they may prioritise speed, secrecy and national advantage over public benefit. If they ignore security, dangerous capabilities may diffuse faster than institutions can respond. A bloom-compatible approach has to hold both sides together: enough control to reduce catastrophic misuse, enough openness and accountability to avoid permanent concentration of power.

Access, legitimacy and public trust

For ordinary people, AI governance will not be judged mainly by summit declarations. It will be judged by whether AI makes schools, hospitals, benefits systems, courts, councils, workplaces and public services fairer or more confusing, more humane or more arbitrary.

Public-sector AI is a crucial test because government decisions carry coercive power. A flawed recommender in a shopping app is one thing; an opaque system affecting welfare payments, immigration, policing, taxation or healthcare access is another. OECD guidance on AI in public service delivery warns that public organisations risk democratic legitimacy if people do not trust AI-enabled services, because experiences with administrative and social services influence trust in government overall.[OECD]one.oecd.orgONE MPCompetition in the Provision of Cloud Computing Services15 May 2025 — Amazon and Microsoft's cloud computing services are reported…Published: May 2025

The UK illustrates both the promise and the problem. The government’s AI Playbook aims to help public bodies use a wider range of AI technologies safely, effectively and responsibly.[GOV.UK]GOV.UKArtificial Intelligence Playbook for the UK GovernmentArtificial Intelligence Playbook for the UK Government Yet civil-society and media scrutiny has repeatedly focused on whether departments disclose algorithmic tools, whether systems are biased, and whether affected people can challenge decisions. Reporting in 2024 found that the UK government had been slow to list AI systems on its transparency register, despite concerns about uses in welfare, immigration and policing.[The Guardian]theguardian.comDespite the development of an algorithmic transparency recording standard in 2021, only nine records have been published so far. The rece…

The lesson is not that public-sector AI should be banned. In a well-governed system, AI could help detect fraud, reduce backlogs, translate services, assist clinicians, triage maintenance, improve accessibility, and give frontline staff better information. The lesson is that legitimacy must be designed in from the start. People need to know when AI is used, what it is used for, what data it relies on, who is accountable, how errors are corrected, and when a human can override the system.

This is especially important because AI systems can shift discretion without making that shift obvious. A caseworker, teacher, doctor or police officer may formally remain “in the loop” while in practice deferring to a model’s ranking, risk score or suggested answer. Research on AI in urban governance argues that AI can reshape both discretion and accountability, creating opportunities for better decision-making but also requiring transparent human-AI collaboration, citizen engagement and robust data governance.[arXiv]arxiv.orgarXiv The rising costs of training frontier AI modelsThe rising costs of training frontier AI modelsMay 31, 2024…Published: May 31, 2024

Trust is not public relations. It is an institutional outcome. It depends on independent audits, appeal rights, procurement competence, published evaluations, incident reporting, whistleblower protection, and the willingness to stop using systems that do not work. A society that wants AI bloom cannot treat the public as a dataset, a market or a passive beneficiary. It has to treat people as rights-bearing participants in decisions that may reshape their lives.

The global gains problem

The AI bloom thesis is universal in language: healthier lives, better science, cleaner energy, more education, less drudgery, a larger long-term future. But the current AI economy is not universal in structure. Compute, cloud infrastructure, frontier labs, technical talent, capital markets and regulatory influence are heavily concentrated in a small number of countries and firms.

That creates a global distribution problem. Countries with weak digital infrastructure, expensive electricity, limited cloud access, few local-language datasets and lower research funding may become consumers of AI systems designed elsewhere. Their citizens may provide data, labour or markets without gaining much control over the tools that shape education, agriculture, public administration, finance and media. UNCTAD’s Technology and Innovation Report 2025 argues for an “AI-for-all” approach that addresses infrastructure, data and skills, and calls for global collaboration to steer AI towards shared goals and values.[UN Trade and Development (UNCTAD]unctad.orgtir2025ch5 entir2025ch5 en

The United Nations has begun building more inclusive AI governance mechanisms. In August 2025, the UN General Assembly established an Independent International Scientific Panel on AI and a Global Dialogue on AI Governance, intended to connect scientific evidence with policymaking and provide an inclusive platform for states and stakeholders.[United Nations]un.orgSource details in endnotes. Chatham House’s assessment is cautious: the UN’s new architecture is mostly powerless in hard enforcement terms, but it could still set agendas and widen participation if implemented well.[Chatham House]chathamhouse.orgcan uns new ai governance efforts weather ai racecan uns new ai governance efforts weather ai race

Benefit-sharing proposals try to make this more concrete. Governance.ai work on international AI benefit sharing describes mechanisms such as sharing AI resources, expanding access to AI systems, and transferring a share of profits or other benefits.[cdn.governance.ai]cdn.governance.aiOpen source on governance.ai. Oxford Martin AI Governance Initiative material frames AI benefit-sharing as a combination of economic redistribution, technology transfer, capacity-building, and safety or non-proliferation controls.[Oxford Martin AIGI]aigi.ox.ac.ukOxford Martin AIGIAI BENEFIT-SHARING FRAMEWORK: BALANCINGOxford Martin AIGIAI BENEFIT-SHARING FRAMEWORK: BALANCING

The difficulty is that these aims can pull against each other. Wider access can accelerate development and help poorer countries solve local problems. But some capabilities may be dangerous if released without safeguards. Technology transfer can build sovereignty, but can also intensify geopolitical rivalry. Profit-sharing can sound fair, but may be hard to measure, easy to capture, or too small to matter. Data dividends can recognise public contribution to AI value, but research on such schemes warns that apparently simple design choices can produce concentrated or demographically uneven payouts.[arXiv]arxiv.orgarXiv The rising costs of training frontier AI modelsThe rising costs of training frontier AI modelsMay 31, 2024…Published: May 31, 2024

A serious global gains agenda therefore has to be plural. It should include affordable access to safe AI tools, local-language datasets, regional compute and data commons, technical training, public-interest research funding, better bargaining power for smaller countries, and rules that stop the most powerful actors from exporting risk while retaining value. The goal is not charity after the fact. It is a world in which more societies can shape the technology before it shapes them.

Institutions for sharing gains safely

The most useful governance ideas are not slogans such as “open” or “closed”, “regulate” or “innovate”. They are institutional designs that answer three questions at once: how to spread benefits, how to prevent capture, and how to manage risk.

One family of institutions concerns access to key inputs. Public compute programmes, national research resources, cloud credits for universities and nonprofits, and regional AI infrastructure can help ensure that scientific and public-interest work is not priced out. They can also support independent safety research, which is hard to do if only frontier companies can afford meaningful experiments. Public compute should be paired with clear allocation criteria, security controls, environmental standards and support for smaller institutions that lack grant-writing or engineering capacity.

A second family concerns market structure. Competition policy can scrutinise cloud lock-in, exclusive partnerships, acquisitions, self-preferencing, restrictive licensing and bundling. The point is not to punish success, but to keep the AI ecosystem contestable enough that new entrants, open-source developers, public-interest labs and sector-specific providers can compete. The European Commission has increasingly linked cloud and AI to Digital Markets Act priorities, while competition scholarship has stressed the importance of access to compute, data and strategic positioning in foundation model competition.[Reuters]reuters.comFollowing positive results in existing digital areas, EU regulators now plan to assess whether cloud providers such as Amazon and Microso…

A third family concerns public procurement. Governments will be major AI customers. If they buy badly, they may entrench vendor lock-in, import bias, weaken public capacity and make critical services dependent on opaque systems. The Ada Lovelace Institute warns that public-sector procurement often lacks the transparency and fairness needed for AI products, and that market concentration and knowledge asymmetries worsen the problem.[Ada Lovelace Institute]adalovelaceinstitute.orgcomputing commonscomputing commons Good procurement should require auditability, interoperability, data protection, clear liability, performance evidence, exit rights, and meaningful consultation with affected workers and service users.

A fourth family concerns frontier safety and accountability. The AI Seoul Summit produced Frontier AI Safety Commitments under which leading AI organisations agreed to publish safety frameworks and address severe risks.[GOV.UK]GOV.UKfrontier ai safety commitments ai seoul summit 2024frontier ai safety commitments ai seoul summit 2024 Such commitments are useful early scaffolding, especially where law lags technology, but voluntary promises are not enough for systems that may affect national security, labour markets, public services and global inequality. They need to be backed by independent testing, incident reporting, standards, liability, whistleblower channels and, where necessary, enforceable release gates.

A fifth family concerns direct economic sharing. If AI substantially raises productivity and profits, societies can distribute gains through ordinary fiscal tools as well as AI-specific mechanisms: taxation, social insurance, public investment, worker ownership, wage subsidies, lifelong learning, shorter working time, universal services, or sovereign wealth-style funds. The important point is not to pick one fashionable mechanism too early. It is to recognise that broad benefit will require deliberate bargaining over capital, labour, data, public infrastructure and intellectual property.

The safest sharing regime will not simply maximise access to the most powerful systems. It will tier access by risk. Low-risk tools for education, translation, accessibility, scientific literature review or local administration can be widely diffused. Higher-risk capabilities involving cyber operations, biological design, autonomous weapons or large-scale persuasion may need stricter controls. Bloom requires both generosity and restraint.

What could go wrong even with good intentions?

Governance can fail in several different ways, and each failure would produce a different kind of unequal future.

The first failure is capture. AI policy may be written through close relationships between governments and the firms they depend on for expertise, infrastructure and economic growth. That can lead to weak oversight, public subsidies without public value, and rules that smaller competitors cannot afford to satisfy.

The second failure is symbolic governance. Governments may publish principles, ethics frameworks and voluntary codes while avoiding hard questions about enforcement, procurement, liability, competition and public capacity. The World Benchmarking Alliance has warned that many leading technology companies disclose high-level ethical AI principles without comprehensive human rights impact assessments, which illustrates the gap between commitments and proof.[Reuters]reuters.comOpen source on reuters.com.

The third failure is over-centralisation in the name of safety. Some frontier capabilities genuinely require control, but safety arguments can also be used to protect incumbents, restrict open research, or concentrate decision-making in a small alliance of states and firms. A world where only a few actors can build or inspect powerful AI may be safer in some respects and more dangerous in others, because mistakes, abuses or strategic decisions by those actors become civilisation-scale bottlenecks.

The fourth failure is under-governance in the name of innovation. If companies and states race to deploy systems without adequate testing, workers may be displaced without support, public services may become less accountable, and military or cyber risks may accelerate. The bloom case depends on long-term trust; repeated scandals could create a backlash that slows beneficial uses while leaving the most powerful private and military deployments untouched.

The fifth failure is global exclusion. If AI governance is negotiated mainly by wealthy countries and frontier companies, poorer countries may be asked to accept standards, tools and dependencies they had little role in shaping. That would weaken legitimacy and could make AI systems less safe, because local knowledge about language, culture, informal economies, public institutions and social risk would be missing.

These risks do not cancel the optimistic case for AI. They make it more demanding. The question is whether societies can build institutions fast enough, competent enough and legitimate enough to channel a powerful general-purpose technology towards broad human flourishing.

What would a fair AI-enabled future look like?

A fair AI-enabled future would not mean that every person, company or country controls identical systems. Some concentration is likely, especially at the frontier. Some restrictions are justified, especially for dangerous capabilities. Some rewards should flow to those who take real risks, build useful products and invest in infrastructure. Fairness does not require flattening every difference.

It does require that the basic direction of the technology is not set solely by private profit, military competition or the preferences of already rich users. In a bloom-compatible future, the gains from AI would show up as widely felt improvements: faster medical discovery, cheaper expertise, better public services, accessible education, safer work, stronger scientific institutions, more capable poorer countries, and greater resilience against global risks. People would not need to understand every model parameter to know who is accountable, how to challenge mistakes, and why the system serves public value.

The practical test is whether AI expands agency. Do workers have a voice in how systems are introduced? Can patients and clinicians question medical AI? Can teachers adapt tools rather than being replaced by rigid platforms? Can public agencies audit and exit contracts? Can smaller countries build local capacity? Can researchers outside frontier companies study risks? Can citizens see how AI is used in decisions that affect them? Can democratic institutions say no to deployments that are profitable but harmful?

AI bloom is sometimes imagined as a technological threshold: build sufficiently powerful intelligence and abundance follows. Governance, power and distribution suggest a different picture. Bloom is not just what advanced AI can do. It is what humans choose to let it become, who gets to participate in that choice, and whether the institutions around it are strong enough to turn capability into shared flourishing rather than concentrated control.

Amazon book picks

Further Reading

Books and field guides related to Who Owns an AI Enabled Future?. Use these as the next step if you want deeper reading beyond the article.

Book

Atlas of AI

By Kate Crawford

Focuses on power, infrastructure and political economy of AI.

See on Amazon

Book

The Coming Wave

By Mustafa Suleyman

Addresses control, concentration and governance of transformative technologies.

See on Amazon

Book

Power and Progress

By Daron Acemoglu, Simon Johnson

Explores who benefits from technological change and how institutions shape outcomes.

See on Amazon

Book

The Age of A.I.: And Our Human Future

By Henry Kissinger

Examines governance and geopolitical implications of AI.

See on Amazon

Browse more on Amazon:Atlas of AIThe Coming WavePower and Progress

As an Amazon Associate I earn from qualifying purchases.

eBay marketplace picks

Marketplace Samples

Example marketplace items related to this page. Use the search link to explore similar finds on eBay.

Shop location

UsingUSA USA USAUKCanadaAustraliaIreland

Example eBay listing

Artificial Intelligence Movie Poster A1 A2 A3

SearcheBay.co.uk: artificial intelligence poster

Browse similar oneBay.co.uk

Example eBay listing

A.I. Artificial Intelligence - Jude Law - One Sheet Cinema Poster

SearcheBay.co.uk: artificial intelligence poster

Browse similar oneBay.co.uk

Example eBay listing

A.i. ARTIFICIAL INTELLIGENCE (2001) UK cinema poster Spielberg Kubrick Jude Law

SearcheBay.co.uk: artificial intelligence poster

Browse similar oneBay.co.uk

Example eBay listing

AI ARTIFICIAL INTELLIGENCE . 2001 ORIGINAL MOVIE POSTER vintage 24 YEARS OLD

SearcheBay.co.uk: artificial intelligence poster

Browse similar oneBay.co.uk

Browse more oneBay.co.uk

Example items shown for inspiration; availability and pricing can change. Branchoria may earn a commission if you purchase through outbound eBay links.

Endnotes

1. Source: arxiv.org
Title: arXiv The rising costs of training frontier AI models
Link:https://arxiv.org/abs/2405.21015

Published: May 31, 2024

2. Source: GOV.UK
Title: Cloud services market investigation
Link:https://www.gov.uk/cma-cases/cloud-services-market-investigation

Published: October 5, 2023

3. Source: one.oecd.org
Link:https://one.oecd.org/document/DAF/COMP%282025%298/en/pdf

Published: May 2025

4. Source: arxiv.org
Title: arXiv Market Concentration Implications of Foundation Models
Link:https://arxiv.org/abs/2311.01550

5. Source: nsf.gov
Link:https://www.nsf.gov/focus-areas/ai/nairr

6. Source: nsf.gov
Title: nairr oc foundations operating national artificial intelligence
Link:https://www.nsf.gov/funding/opportunities/nairr-oc-foundations-operating-national-artificial-intelligence

7. Source: GOV.UK
Title: A I Research Resource
Link:https://www.gov.uk/government/publications/ai-research-resource

8. Source: 2021-2025.state.gov
Link:https://2021-2025.state.gov/bureau-of-arms-control-deterrence-and-stability/political-declaration-on-responsible-military-use-of-artificial-intelligence-and-autonomy/

9. Source: sipri.org
Title: 12. Artificial intelligence and international peace and security
Link:https://www.sipri.org/sites/default/files/SIPRIYB25c12.pdf

10. Source: oecd.org
Title: ai in public service design and delivery 09704c1a
Link:https://www.oecd.org/en/publications/2025/06/governing-with-artificial-intelligence_398fa287/full-report/ai-in-public-service-design-and-delivery_09704c1a.html

11. Source: GOV.UK
Title: Artificial Intelligence Playbook for the UK Government
Link:https://www.gov.uk/government/publications/ai-playbook-for-the-uk-government/artificial-intelligence-playbook-for-the-uk-government-html

12. Source: arxiv.org
Link:https://arxiv.org/abs/2502.13101

13. Source: unctad.org
Title: tir2025ch5 en
Link:https://unctad.org/system/files/official-document/tir2025ch5_en.pdf

14. Source: unctad.org
Title: tir2025 en
Link:https://unctad.org/system/files/official-document/tir2025_en.pdf

15. Source: cdn.governance.ai
Link:https://cdn.governance.ai/Options_and_Motivations_for_International_AI_Benefit_Sharing.pdf

16. Source: arxiv.org
Link:https://arxiv.org/abs/1912.00757

17. Source: reuters.com
Link:https://www.reuters.com/legal/litigation/eu-rules-reining-big-tech-will-now-target-cloud-services-ai-regulators-say-2026-04-28/

18. Source: GOV.UK
Title: frontier ai safety commitments ai seoul summit 2024
Link:https://www.gov.uk/government/publications/frontier-ai-safety-commitments-ai-seoul-summit-2024

19. Source: reuters.com
Link:https://www.reuters.com/technology/global-ai-summit-seoul-aims-forge-new-regulatory-agreements-2024-05-21/

20. Source: reuters.com
Link:https://www.reuters.com/sustainability/boards-policy-regulation/with-ai-accountability-stalling-boards-must-push-tech-giants-greater–ecmii-2026-02-04/

21. Source: oecd.org
Title: 795de142 en
Link:https://www.oecd.org/content/dam/oecd/en/publications/reports/2025/06/governing-with-artificial-intelligence_398fa287/795de142-en.pdf

22. Source: oecd.org
Title: governing with artificial intelligence 398fa287
Link:https://www.oecd.org/en/publications/2025/06/governing-with-artificial-intelligence_398fa287.html

23. Source: oecd.org
Link:https://www.oecd.org/en/publications/2025/06/governing-with-artificial-intelligence_398fa287/full-report/how-artificial-intelligence-is-accelerating-the-digital-government-journey_d9552dc7.html

24. Source: oecd.org
Link:https://www.oecd.org/en.html

25. Source: oecd.org
Title: component 6
Link:https://www.oecd.org/en/publications/competition-in-artificial-intelligence-infrastructure_623d1874-en/full-report/component-6.html

26. Source: oecd.org
Title: artificial intelligence
Link:https://www.oecd.org/en/topics/policy-issues/artificial-intelligence.html

27. Source: oecd.org
Title: ai in policy evaluation c88cc2fd
Link:https://www.oecd.org/en/publications/2025/06/governing-with-artificial-intelligence_398fa287/full-report/ai-in-policy-evaluation_c88cc2fd.html

28. Source: one.oecd.org
Link:https://one.oecd.org/document/DAF/COMP/GF%282025%294/en/pdf

29. Source: oecd.org
Title: full report
Link:https://www.oecd.org/en/publications/2025/06/governing-with-artificial-intelligence_398fa287/full-report.html

30. Source: oecd.org
Link:https://www.oecd.org/content/dam/oecd/en/publications/reports/2026/02/oecd-ai-observatory-index_8f5fa0f2/32c01014-en.pdf

31. Source: oecd.org
Link:https://www.oecd.org/en/topics/sub-issues/ai-compute.html

32. Source: one.oecd.org
Link:https://one.oecd.org/document/DAF/COMP/GF%282025%291/en/pdf

33. Source: one.oecd.org
Link:https://one.oecd.org/document/COM/DSTI/CDEP/STP/GOV/PGC%282024%291/FINAL/en/pdf

34. Source: cdn.governance.ai
Title: Market Concentration Implications of Foundation Models
Link:https://cdn.governance.ai/Market_Concentration_Implications_of_Foundation_Models.pdf

35. Source: GOV.UK
Title: ai foundation models initial report
Link:https://www.gov.uk/government/publications/ai-foundation-models-initial-report

36. Source: GOV.UK
Title: ai opportunities action plan
Link:https://www.gov.uk/government/publications/ai-opportunities-action-plan/ai-opportunities-action-plan

37. Source: GOV.UK
Title: frontier ai safety commitments ai seoul summit 2024
Link:https://www.gov.uk/government/publications/frontier-ai-safety-commitments-ai-seoul-summit-2024/frontier-ai-safety-commitments-ai-seoul-summit-2024

38. Source: assets.publishing.service.gov.uk
Title: publishing.service.gov.uk A I Foundation Models
Link:https://assets.publishing.service.gov.uk/media/661e5a4c7469198185bd3d62/AI_Foundation_Models_technical_update_report.pdf

39. Source: assets.publishing.service.gov.uk
Title: Full report
Link:https://assets.publishing.service.gov.uk/media/65081d3aa41cc300145612c0/Full_report_.pdf

40. Source: assets.publishing.service.gov.uk
Link:https://assets.publishing.service.gov.uk/media/65cb77caa7ded0000c79e526/Government_response_to_the_House_of_Lords_AI_in_Weapon_Systems_Committee_Report.pdf

41. Source: united.com
Link:https://www.united.com/

42. Source: microsoft.com
Link:https://www.microsoft.com/en-us/research/project/national-ai-research-resource-nairr-pilot/

43. Source: oecd.ai
Title: can mid sized economies come together to build frontier ai
Link:https://oecd.ai/en/wonk/can-mid-sized-economies-come-together-to-build-frontier-ai

44. Source: oecd.ai
Title: the geopgraphy of ai compute mapping what is available and where
Link:https://oecd.ai/en/wonk/the-geopgraphy-of-ai-compute-mapping-what-is-available-and-where

45. Source: oecd.ai
Link:https://oecd.ai/en/catalogue/tools/a-frontier-ai-risk-management-framework-bridging-the-gap-between-current-ai-practices-and-established-risk-management

46. Source: cloud.google.com
Title: idc marketscape 2025 ww foundation models
Link:https://cloud.google.com/resources/content/idc-marketscape-2025-ww-foundation-models

47. Source: economic-policy.org
Title: EcPol 2023 183.R1 Proof hi Korinek Vipra
Link:https://www.economic-policy.org/wp-content/uploads/2024/03/EcPol-2023-183.R1_Proof_hi_Korinek_Vipra.pdf

48. Source: arxiv.org
Link:https://arxiv.org/pdf/2510.15200

49. Source: industry.gov.au
Title: seoul declaration countries attending ai seoul summit 21 22 may 2024
Link:https://www.industry.gov.au/publications/seoul-declaration-countries-attending-ai-seoul-summit-21-22-may-2024
Published: may 2024

50. Source: newsletter.safe.ai
Title: ai safety newsletter 43 white house
Link:https://newsletter.safe.ai/p/ai-safety-newsletter-43-white-house

51. Source: adalovelaceinstitute.org
Title: computing commons
Link:https://www.adalovelaceinstitute.org/report/computing-commons/

52. Source: europarl.europa.eu
Title: European Parliament Defence and artificial intelligence
Link:https://www.europarl.europa.eu/RegData/etudes/BRIE/2025/769580/EPRS_BRI%282025%29769580_EN.pdf

53. Source: Lawfare
Title: to govern ai we must govern compute
Link:https://www.lawfaremedia.org/article/to-govern-ai-we-must-govern-compute

54. Source: theguardian.com
Link:https://www.theguardian.com/technology/article/2024/aug/25/register-aims-to-quash-fears-over-racist-and-biased-ai-tools-used-on-uk-public

55. Source: theguardian.com
Link:https://www.theguardian.com/technology/2024/nov/28/uk-government-failing-to-list-use-of-ai-on-mandatory-register

56. Source: un.org
Link:https://www.un.org/global-digital-compact/en/ai

57. Source: chathamhouse.org
Title: can uns new ai governance efforts weather ai race
Link:https://www.chathamhouse.org/2025/09/can-uns-new-ai-governance-efforts-weather-ai-race

58. Source: aigi.ox.ac.uk
Title: Oxford Martin AIGIAI BENEFIT-SHARING FRAMEWORK: BALANCING
Link:https://aigi.ox.ac.uk/wp-content/uploads/2025/12/Operationalising-AI-Benefit-Sharing-Workshop-02.12.2025.pdf

59. Source: adalovelaceinstitute.org
Title: public sector ai
Link:https://www.adalovelaceinstitute.org/policy-briefing/public-sector-ai/

60. Source: unsceb.org
Title: United Nations System White Paper on AI Governance
Link:https://unsceb.org/sites/default/files/2024-04/United%20Nations%20System%20White%20Paper%20on%20AI%20Governance.pdf

61. Source: chathamhouse.org
Title: ai export controls are not best bargaining chip
Link:https://www.chathamhouse.org/2026/04/ai-export-controls-are-not-best-bargaining-chip

62. Source: techpolicyinstitute.org
Title: AI Governance
Link:https://techpolicyinstitute.org/wp-content/uploads/2026/03/AI_Governance.pdf

Additional References

63. Source: youtube.com
Link:https://www.youtube.com/watch?v=Hw1ylGymNTE

64. Source: youtube.com
Title: Who Controls Your AI Agents? Responsible AI Governance
Link:https://www.youtube.com/watch?v=Sn2R5UgvnKg

65. Source: media.defense.gov
Link:https://media.defense.gov/2025/Dec/23/2003849070/-1/-1/1/ANNUAL-REPORT-TO-CONGRESS-MILITARY-AND-SECURITY-DEVELOPMENTS-INVOLVING-THE-PEOPLES-REPUBLIC-OF-CHINA-2025.PDF

66. Source: youtube.com
Title: Who Controls the Future of AI: The Oligarchs or the People?
Link:https://www.youtube.com/watch?v=pJ-PlOCpkyY

67. Source: youtube.com
Title: New frontiers in AI governance | Nitarshan Rajkumar
Link:https://www.youtube.com/watch?v=BgKHMwc-YPk

68. Source: researchgate.net
Link:https://www.researchgate.net/publication/391396519_Bridging_the_International_AI_Governance_Divide_Key_Strategies_for_Including_the_Global_South

69. Source: researchgate.net
Link:https://www.researchgate.net/publication/391209292_Global_AI_Benefit-Sharing_Mechanisms_Governance_Levers_and_Implementation_Pathways

70. Source: cerre.eu
Link:https://cerre.eu/wp-content/uploads/2025/06/A-Competition-Policy-for-Cloud-and-AI_FINAL.pdf

71. Source: cigionline.org
Link:https://www.cigionline.org/documents/3745/AI_National_Security.pdf

72. Source: cigionline.org
Link:https://www.cigionline.org/documents/3748/Policy_Brief_No._225_Maheshwari.pdf