In an August 2024 working paper, Joseph Stiglitz identifies the many failures of neoliberal ideology, going beyond the economic; a polarized society, loss of well-being, and a degraded environment. Stiglitz notes that a reckless “privatization of gains” and “socialization of losses” under neoliberalism reflects an exploitative society in which the powerful rewrite the rules of the economy to serve their interests.

The institutional design of digital society is unfortunately premised on the failed political ideology of neoliberalism. It has destroyed, what Stiglitz terms, “sustainable democracy.” Market-led digitalization has neither produced trickle-down growth nor succeeded in self-governance. This is why the UN Global Digital Compact (GDC) negotiations (currently at Rev 4.0) provoke a sinking deja vu.

The Compact has very lofty ambitions as evidenced by its Preamble: bridging the digital divides between and within states and advancing an equitable digital environment for all (para 8a); furthering development-oriented digital cooperation rooted in the 2030 Agenda that eradicates poverty and leaves no one behind (para 8b); advancing all human rights, including the right to development (para 8c); and enabling equitable and meaningful inclusion in the digital economy by tackling existing concentrations of technological capacity and market power (para 8f). Unfortunately, the ambition is not matched by corresponding commitments to transform the business-as-usual macroeconomics that has reinforced global digital inequality. The picture is bleaker when we consider how negotiations on the Pact for the Future (as part of which the Global Digital Compact is being discussed) to reinvigorate the multilateral system also do little to shift the global development status quo. Blindsiding the self-serving rules crafted by the powerful, the emerging text in these negotiations seems to be content with a formulaic call-back to market anachronisms.

Under the circumstances, and without “a post-neoliberal macroeconomic policy framework” (quoting Stiglitz), a just and sustainable digital transition cannot be contemplated. We write to express this caution – questioning the GDC’s adherence to neoliberal dogma that advocates for private sector investment as the holy grail for digital infrastructure development; interoperability in data governance as the magic bullet for unlocking the benefits of data; a human rights frame oblivious to the injustices of the digital economy; and a laissez-faire multistakeholderism for digital governance divested of accountability.

In the last stages of the negotiations, outcomes for the global majority will depend entirely on concrete prescriptions to tackle the historical power structures that have shaped the digital. We, therefore, urge the negotiators of the GDC not to miss the woods for the trees. An equitable and just digital future hinges on a GDC that includes the following non-negotiables:

1. International public financing commitments for digital infrastructure development in the Global South. The role of Official Development Assistance (ODA) and new mechanisms like the Digital Development Tax proposed by the UN Secretary-General and the global financial transactions tax proposed by civil society groups cannot be overemphasized.
2. Recognition of the strategic autonomy of member states to determine their independent approaches to the governance of cross-border data flows. ‘Data flows with trust’ cannot become the universal prescription for an innovation economy. An equitable digital future begins with policy autonomy and not neoliberal openness.
3. Concrete commitments on labor rights, environmental rights, and the right of all to benefit from scientific endeavor, in data and artificial intelligence (AI) value chains. Reducing human rights to privacy and individual freedoms, without addressing the injustices of trade and intellectual property (IP) regimes in the AI economy, only perpetuates a neo-colonial economic order.
4. Clear guidance for ensuring outcome legitimacy in future global digital cooperation arrangements on internet, data, and AI governance. Mere lip service to the World Summit on the Information Society (WSIS) consensus with no direction on outstanding and emerging issues in digital public policy will only further the elite capture of these arrangements. Unless the GDC spells out the guiding norms for representational legitimacy and the specific issue-based configuration to determine the relevant stakeholders, new mechanisms in the GDC risk undercutting public interest.

Blindspot 1. Private sector investment as the holy grail for digital infrastructure development

The GDC identifies private sector investment as the most significant contributor to bridging digital infrastructure deficits in the South. A range of options are proposed in this regard: blended financing for affordable connectivity models, multistakeholder partnerships for digital public goods and infrastructure in critical development domains, and innovative financing options involving the private sector and philanthropy for AI capacity-building. These solutions ignore lessons from past reliance on market mechanisms to close the access divide in internet connectivity (See Box 1).

Just as in the case of connectivity, in AI infrastructure as well there is growing evidence that market-based investments do not guarantee democratization of benefits. Venture capital funding in AI ecosystems not only prioritizes dominant actors and incumbent monopolies but also results in hasty divestment from local companies building contextually grounded AI models. The result is that communities in the Global South are subject to a double whammy: subpar products created by Big Tech companies that don’t consider them a priority, and elimination of local alternatives due to the hype that powerful companies create. Recently, Meta claimed to have an AI language model that provided state-of-the-art machine translation across 200 languages, including 54 African languages. The model was found to have many quality issues for languages such as Amharic, Tigrina, and two dialects of Twi. In these specific languages, local organizations like Ghana NLP and Lesan outperformed Meta’s products. However, Meta’s model resulted in driving out investment from local businesses.

The evidence is unmistakable; a market-first approach to public digital infrastructure will not result in real inclusion or promote diverse, locally relevant innovation.

Relying on the market and private philanthropy to fill core infrastructural deficits can bring graver threats. Innovation capability in the AI ecosystem is founded in compute power – the stack of hardware, software, and other infrastructure components that generate computational processing capacity for AI model-building. As investments in compute power are prohibitively expensive, we find ourselves in a scenario where “only Big Tech has the resources to upgrade their compute capacity while simultaneously being able to collect data to train ML/DL (machine learning/deep learning) models and to hire the specialized AI talent to work on these models.” Big Tech dominance has skewed the directions of AI research towards “data-hungry and computationally intensive deep learning methods” at the “expense of research involving other AI methods, research that considers the societal and ethical implications of AI, and applications in sectors like health.” Recovering the space for smaller firms and non-elite academic centers to participate in AI development can bring more diversity to future innovation trajectories, including the exploration of smaller models that are attentive to data extractivism and attendant threats to ecological and social systems.

The evidence is unmistakable; a market-first approach to public digital infrastructure (ranging from connectivity to digital public goods, data and statistical capability, compute infrastructure, etc.) will not result in real inclusion or promote diverse, locally relevant innovation. On the contrary, it can shrink the space for the market of ideas, resulting in a capture by few actors. What we need therefore are public financing strategies for new pathways to digital industrialization in the Global South, including dedicated investments for a public compute infrastructure development strategy.

The Digital Development Tax is a useful and timely idea, considering recent developments in global taxation debates, including in the G20. The Digital Development Tax was proposed by the UN Secretary-General as a mandatory contribution from transnational platform companies who have profited from the internet to close the connectivity gap. The global financial transactions tax proposed by civil society groups is another option to consider.

The creation of foundational AI models for public good cannot be left to Big Tech whose business models are based on the extraction of public data without any reciprocal commitment to share benefits from such data. Only public funding can advance the collaborative development of foundational AI models for the public good, “taking into consideration the environmental, social and cultural impacts of AI, free from the imperative to make a profit.” Towards this, an effective route may be the creation of an international research institution modeled after CERN and “as with the actual CERN’s particle accelerators and supercomputers, a CERN-like body for AI would be equipped with its own infrastructure to support computational research.”

Blindspot 2. Conflating technical interoperability with the political governance of data

Promoting interoperable data governance approaches is a major objective of the GDC. Interoperability is a term that originated in technical literature to describe the ability of two or more digital systems to work together. Data interoperability, as technical terminology, refers to the use of common data formats and protocols that enable information technology systems to communicate with each other. Maintaining technical interoperability is necessary and useful for datasets and data regimes to be compatible with each other. Yet, extending the minimum technical interoperability approach to the entirety of data governance is fraught with risks. Data governance, as a systemic framework, maybe described as a techno-political regime, incorporating constituent elements of “data handling rules, consumer rights, oversight institutions, and enforcement mechanisms that jointly enable the safe and trustworthy exchange of data flows across jurisdictions.” The technical interoperability of data is hence not the same as the complex set of policies comprising the data governance regime aimed at nurturing just and equitable digital ecosystems.

The GDC’s exhortation for “innovative, interoperable and inclusive mechanisms to enable data to flow with trust within and between countries to mutual benefit, while respecting relevant data protection and privacy safeguards and applicable legal frameworks” (para 46), is commonly heard in the dominant data governance discourse. The G20 Osaka Declaration also emphasizes the need to advance interoperability of data governance to aid data free flows with trust and harness the potential of data for economic opportunity. Similarly, the UN High-Level Advisory Board on Effective Multilateralism highlights that in the current context of poor interoperability, a system for trusted and secure data flows is urgently needed.

The distinction between data interoperability and data governance interoperability is extremely material to global digital justice. Most discussions of data governance interoperability or data flows with trust focus on legal mechanisms for the transfers of personal data with some safeguards. Anchoring everyone to an acceptable benchmark in consumer rights and privacy protections can drive standards to the lowest common denominator and ignore a whole gamut of data rights. Preserving and promoting data rights in data flows is about addressing the political economy of the platform marketplace – who gains and who loses as data crosses borders in new-age value chains. Every country needs to carefully consider its strategic interests in different sectors and domains that may be best served through cross-border data flows. There is no monolithic roadmap.

As such, global data interoperability in a winner-takes-all market model comes with a huge disadvantage for developing countries. As UNCTAD’s Digital Economy Report 2021 (p.131) argues, “While access to data is a necessary condition to benefit from data, it is not sufficient.” The economic benefits of data flows are not evenly distributed among countries. In the absence of a globally accepted framework for valuing data resources, developing countries have no recourse for a fair share in the gains of the digital economy. Without the capabilities for data-based innovation, they are in the inevitable position of importing digital products and services – built on the exodus of their own data resources – at exorbitant costs. Data resources, in and of themselves, also do not fetch significant export earnings. The GDC appears to be reconciled to this unequal distribution of data capabilities in shaping the trajectories of future innovation. Para 53’s (Rev 4) strategy for international cooperation to build the AI capacities of developing countries is limited to the import and customization of globally developed AI models through the use of locally generated data in their application.

Every country needs to carefully consider its strategic interests in different sectors and domains that may be best served through cross-border data flows. There is no monolithic roadmap.

Data innovation tends to consolidate knowledge monopolies in different sectors. Even where global governance regimes lay down clear rules, such as in the biodiversity domain, interoperability of data can give rise to a new generation of problems that erode the rights of developing countries as well as the public benefits of data innovation (See Box 2).

Pushing for interoperability as a foundational norm in data governance in the current status quo risks reinforcing the ills of neoliberal globalization, ignoring the political, cultural, and economic realities of particular contexts. UNCTAD’s survey of the data governance frameworks of G20 countries reveals how countries approach the definition of sensitive data very differently. This can be traced to national differences in fundamental beliefs and values about privacy, public interest, and commercialization of information. Imposing a dominant regulatory grammar of data governance on the majority world amounts to a new form of imperialism that takes away the autonomy of a country to independently determine the overall goal of its digital policy, including the appropriate politico-economic balance it seeks between privacy protection, data security, public innovation, private enterprise, etc., in the digital society and economy. For example, critics have highlighted how the policy push for mirroring the EU’s General Data Protection Regulation (GDPR) standards in domestic data protection regulation in African countries without concern for their “unique social values and economic realities” may jeopardize the emergence of dynamic, home-grown alternatives to development.

The road to interoperable data governance must focus on sustainable, inclusive, and democratic futures for all.

Governance regimes for cross-border data flows need more than technical measures. They need robust legal and institutional mechanisms so that data rights of peoples and nations – encompassing much more than personal data protection – can be upheld through enforceability and redress options. The road to interoperable data governance must focus on sustainable, inclusive, and democratic futures for all.

The GDC should push for an accountable, inclusive, and equitable data governance approach rooted in distributive and democratic integrity. Systematic and inclusive deliberations are needed to move beyond technical interoperability rules in data. A comprehensive global data governance must concern itself with both meta norms for a brave, new data/AI world grounded in digital justice as well as the specific principles, rules, and institutional frameworks reflecting our datafied reality for different international and national policy domains.

Blindspot 3. Human rights sans structural justice

The concrete action commitments in the Human Rights section of the GDC Rev 4 are limited to protecting users from “violations, abuses and all forms of discrimination” through the lifecycle of digital and emerging technologies (para 22). Such a narrow approach focused on individual rights violations in the use and downstream impact of technological products falls short for the following reasons.

The human rights section of the GDC should contain explicit commitments to address labor and environmental rights violations in AI value chains.

Firstly, it ignores the gross violations of labor and environmental rights in the digital supply chains that are integral to AI production. AI models depend on an army of ghost workers from the Global South engaged in data labeling, annotation, and content moderation tasks without minimum wage guarantees or any form of labor protection. Precarious workers in countries such as India, Kenya, the Philippines, and Mexico work in nearly sweatshop-like conditions as part of the value chain of companies such as Meta and Open AI. Data-intensive AI models also have a huge ecological footprint – from increasing greenhouse gas emissions and water consumption of data centers to hazardous waste generated in the production of semi-conductor chips that are an essential component of physical compute infrastructure. Evidence suggests that the social harms of environmental and labor exploitation multiply in corporate-controlled food chains deploying farm-to-fork platform solutions, directly impinging on the rights of marginal and small peasants and cultivators. AI-supported prospecting by Big Oil enables an intensification of fossil fuel extraction that undercuts the spirit of climate change mitigation. After a Greenpeace exposé in 2020, some Big Tech firms came out with public announcements that they would stop making AI tools for oil and gas exploration, but pronouncements by Big Tech for ethical conduct have historically not added up to much. On the contrary, Big Tech companies have increasingly sought to rewrite the rules on net zero and abused the notion of sustainability in AI technologies.

Secondly, reducing human rights to negative rights fails to address the structural exclusion of the majority world from the benefits of AI development. The positive right of all people to share in the benefits of scientific progress is core to digital human rights. Today’s innovation ecosystem is controlled by Big Tech firms whose business model is predicated on the accumulation of data-driven innovation rents accruing through data centralization and monopoly capture of downstream benefits in AI value chains. Big Tech also strategically leverages existing global trade and intellectual property rights (IPR) regimes to consolidate intellectual monopolies. By pursuing aggressive patenting strategies that squelch competition in AI innovation, and lobbying for digital trade policy rules that prevent governments in the Global South from having access to algorithmic source code for public interest intervention, Big Tech has systematically eroded institutional norms and rules.

The human rights section of the GDC should contain explicit commitments to address labor and environmental rights violations in AI value chains. Further, in its section on ‘Expanding Inclusion in and Benefits from the Digital Economy for All,’ the Compact should commit to addressing the barriers posed by current trade and IPR frameworks to the realization of the right to science (including the benefits of AI innovation) by the majority world while also underscoring the acquisition of data and AI capabilities as integral to the right to development. Only then can the vision of human rights as an integrated and indivisible whole – outlined in the Principes section of the GDC – be achieved in practice.

Blindspot 4. Laissez-faire multistakeholderism as digital democracy’s fait accompli

The WSIS enshrined multistakeholderism as a horizontal principle for all elements of internet governance. It also recommended the operationalization of this principle through the establishment of two mechanisms: (a) the Internet Governance Forum (IGF), an annual forum for multistakeholder policy dialogue; and (b) the initiation of a process of ‘enhanced cooperation’ by the UN Secretary-General – to enable governments, on an equal footing along with other stakeholders in their respective roles and responsibilities, to exercise their legitimate authority in the evolution of international public policy issues pertaining to the internet.

Without any oversight function, and constituted as a neutral, non-duplicative, and non-binding process, (Para 77, Tunis Agenda), the IGF has been a mixed bag; acting as a space for multistakeholder dialogue, but lacking the legitimate mechanisms for generating mandates on cross-cutting internet-related public policy issues. On the agenda of ‘enhanced cooperation,’ two multistakeholder working groups were set up, but disbanded without any consensus on a common set of ground norms to guide internet-related public policy. Some commentators have argued that the failure of these working groups is not surprising, considering that the internet governance status quo has always suited US strategic interests propping up its tech behemoths in global digital markets. The absence of international norms for digital public policies has also enabled forum shopping by powerful countries. For instance, for a very long time, the US treated data governance as a trade policy issue, till it made a U-turn in late 2023.

The High-Level Panel on Digital Cooperation set up by the UN Secretary-General in July 2018 pointed to the need for a strengthened multilateralism complemented by multistakeholderism, involving civil society, academics, technologists, and the private sector, particularly from developing countries and traditionally marginalized groups to achieve effective digital cooperation. It put forth a model for a digital commons architecture that would be governed through a model with mutually dependent multilateral and multistakeholder tracks.

The Panel’s report underscored an important principle: that there can be no one-size-fits-all enhanced cooperation mechanism; no singularly perfect complement of multilateralism and multistakeholderism for any and every digital policy issue. The NETmundial 2014 statement also recognizes that “the respective roles and responsibilities of stakeholders should [only] be interpreted in a flexible manner with reference to the issue under discussion.” The devil is always in the details. The principle of how legitimacy is achieved in the governance of internet resources within internet institutions cannot be thoughtlessly applied out of context to all policy scenarios. Multistakeholderism is not a call for “everybody (to) be represented everywhere all at once” in digital policymaking.

The GDC’s action commitments unfortunately hit a dead-end on the specific complementarities between multilateralism with multistakeholderism. To begin with, the GDC does not fix the shortcomings of the IGF that prevent it from emerging as a truly participatory and representative space for multistakeholder dialogue. The provision of voluntary funding (para 29b) is not enough to make diversity and inclusion of governments and other stakeholders from developing countries a reality in the space.

Secondly, though the GDC reaffirms that “Internet governance should continue to follow the provisions set forth in the outcomes of the summits held in Geneva and Tunis, including in relation to enhanced cooperation” (para 27, Rev 4), there is no clarity on the concrete agenda to be addressed in this regard. The action commitments in the GDC’s section on internet governance, for instance, do not acknowledge the unfinished agenda of internet-related public policy issues that have been open for over 20 years after the WSIS. In 2005, the Working Group on Internet Governance (WGIG) – a multistakeholder body entrusted by the WSIS-Phase 1 to investigate and make proposals for action on the governance of the internet to WSIS-Phase 2 – identified several policy issues – from IPR to data protection and privacy rights, freedom of expression, and consumer rights. Sadly, an effective public policy response at the multilateral level is still lacking on these.

Thirdly, in the GDC’s proposals for data and AI governance, it is unclear how a “multilateral, transparent and democratic” process of digital cooperation envisioned in the Tunis Agenda (Para 29) will materialize. A CSTD-led dedicated working group is expected to come up with recommendations on data governance at multiple levels as related to development, including cross-border data flows, on the basis of multistakeholder dialogue, and place them before the UNGA. It remains to be seen how this working group and its multistakeholder deliberation will be configured. What is worrisome is that Big Tech – despite its utter disregard for human rights, democratic integrity, and ecological wellbeing – may well claim its ‘legitimate stake’ at the discussion table, given that the GDC chooses to remain silent on how representational legitimacy will be achieved.

That post-WSIS multistakeholderism has done little to challenge the digital governance status quo with disastrous consequences for economic and social justice is a truism that is widely accepted today.

With respect to the International Scientific Panel on AI, the GDC prescribes that two General Assembly co-facilitators, one from a developed country and one from a developing country, will draw up the terms of reference and modalities through an intergovernmental process and in consultations with other relevant stakeholders. The Compact remains dangerously silent on the relationship of the multistakeholder consultations with the intergovernmental process. It is anybody’s guess how competing interests that emerge in these multistakeholder consultations will be balanced to draw up the terms of reference.

In the final analysis, the multistakeholderist approach in the Compact carries a distinctly laissez-faire flavor – based on the premise that by merely bringing together international institutions, states, and non-state actors to deliberate on policy issues through open dialogue, a public interest consensus is automatically produced. As Mary Ann Manahan and Madhuresh Kumar highlight, since the turn of the millennium, there have been over 21 global multistakeholder initiatives created in different digital policy areas such as AI, cybersecurity, trade and e-commerce, human rights, etc. In practice, these global multistakeholder initiatives have been dominated by powerful countries and Big Tech corporations, and have only pushed forward neoliberal soft law prescriptions that suit dominant interests in the digital economy, without addressing the roots of global injustice such as the “proprietary exploitation of data.” The aggregation of various interests in the room in the name of an equal-stakes dialogue is not the same thing as bottom-up participation of a plurality of voices. Public policy processes need a defined institutional mechanism to mediate competing interests (stakes) and produce a public interest consensus in the policy process that is transparent and accountable.

That post-WSIS multistakeholderism has done little to challenge the digital governance status quo with disastrous consequences for economic and social justice is a truism that is widely accepted today. In this scenario, the GDC needs to commit to IGF strengthening explicitly, particularly in terms of addressing its shortcomings of representational diversity and lack of dedicated funding. It also needs to re-look at the proposals for data and AI governance and set issue-specific norms for democratic digital cooperation in these areas as envisioned by the WSIS. There is only a hard route to achieve this – going back to the drawing board to arrive at a widely deliberated and inclusive process to evolve a binding human rights-based global digital constitutionalism that can serve people and the planet.

Conclusion

Tracing the checkered history of global digital governance, this commentary points to four gaps that need to be urgently addressed for the GDC to retain its radical edge. We have argued that 1. private finance is not going to redeem the unconnected and that public financing is non-negotiable for an equitable and inclusive digital society; 2. the governance of data flows needs more than technical measures; it is a profoundly political issue calling for new imagination about how decolonial and sustainable economies can be built; 3. the GDC cannot further a development agenda that works for the 99% unless economic and social rights, environmental justice, and the right to development are a real part of the conversation; and 4. without a human rights constitutionalism that enables the development of issue-specific norms and institutional mechanisms for upholding public interest in digital policymaking, harking at multistakeholderism will just end up entrenching corporate agendas in these spaces.

During the pandemic, the world witnessed how reliance on a multistakeholder partnership mechanism – COVAX – to procure COVID vaccines for the Global South undermined the global public good of universal healthcare, instead reinforcing the status-quoist IPR regime that emboldens the global pharma value chain to perpetuate vaccine scarcity. The Covid-19 pandemic also showed us that the marketization and commodification of digital infrastructural goods is a slippery slope; it undercuts social autonomy and enfeebles proactive public policy action.

Just like Big Pharma, Big Tech will go to any length to profiteer in the neoliberal global politico-economic regime. The WGIG Report of 2005, referred above, also pointed to two overarching prerequisites to enhance the legitimacy of internet governance processes: “(a) the effective and meaningful participation of all stakeholders, especially from developing countries, and (b) the building of sufficient capacity in developing countries, in terms of knowledge and of human, financial and technical resources.” These prerequisites elude us even today. The GDC must take note of this crisis of legitimacy at the root of global digital injustice to forge a digital order that can dignify the aspirations of the furthest from the first.

The authors would like to thank Richard Hill, President, Association for Proper Internet Governance, Switzerland, for his review, and inputs to the piece.