Who holds digital power accountable?

 

Digitalization has become a constitutive layer of modern society. Digital platforms, cloud services, algorithms, artificial intelligence, data centers and communication networks shape economic activity, public services, knowledge, political participation and social interaction. Major technology companies consequently exercise economic, technological, normative and epistemic power: they establish rules, control essential infrastructure, moderate public communication and influence access to information, opportunities and services. Network effects, data concentration, vertical integration and dependence on closed ecosystems reinforce this power, often without corresponding democratic legitimacy or effective public accountability[1].

The central challenge is therefore not to obstruct technological innovation, but to embed digital power within a democratic, constitutional and polycentric institutional order. Concentrated private power should not simply be replaced by centralized state control. Instead, power must be dispersed among institutions that can monitor, challenge and correct one another.

Digital power operates at two connected levels. Infrastructural power determines who controls platforms, cloud capacity, app stores, data flows, identity systems, and channels of public visibility. Decision power arises when a particular system helps determine employment, credit, healthcare, education, policing, migration, insurance, social security, or access to justice. The first level requires competition policy, interoperability, public-interest obligations, resilience, and democratic oversight; the second requires explanation, human review, contestability, and remedy. Both require traceable responsibility.

Platform regulation must address structural dependence rather than focusing solely on prices or conventional market competition. Data concentration, network effects, algorithmic control, self-preferencing, vertical integration and barriers to market entry should also be considered[2]. Interoperability, data portability and open standards can reduce user lock-in and enable competition. Where behavioral regulation proves insufficient, functional separation or structural intervention may be justified as a proportionate last resort. A resilient digital ecosystem should also include platform cooperatives, public digital services, open-source projects, digital commons and other alternatives alongside commercial platforms.

Algorithms and artificial intelligence increasingly influence decisions concerning employment, credit, healthcare, education, policing, migration, insurance, social security and justice. These systems are not neutral instruments: their datasets, variables, risk thresholds and optimization goals embody normative choices. When such choices remain opaque, decision-making shifts from publicly accountable institutions to technical systems that citizens may be unable to understand or challenge.

Democratic control therefore requires meaningful explainability, independent audits, bias testing and continuing assessments[3] of social and constitutional effects. Public algorithm registers can clarify which systems are being used, for what purposes and under whose supervision. Public algorithm registers should disclose the responsible authority, purpose, legal basis, affected groups, system provider, risk classification, assessment dates, and routes for complaint. Legitimate security and confidentiality interests may justify limited redactions, but not the concealment of a system’s existence, public purpose, or accountable owner. High-impact applications should undergo prior and continuing assessments covering fundamental rights[4], privacy, equality, autonomy, security and institutional correctability. Decisions affecting rights, health, freedom or livelihood must remain subject to human intervention and effective appeal[5]. Human involvement is meaningful only when the reviewer has sufficient authority, competence, time, information, and independence to depart from the automated output. A nominal human signature that merely confirms the system’s recommendation does not provide effective oversight. Citizen panels and multidisciplinary review bodies can help ensure that decisions about acceptable risks and prohibited applications are treated as democratic choices rather than purely technical matters.

Digital power also affects the epistemic foundations of democracy. Recommendation algorithms increasingly determine which information becomes visible, while commercial attention models tend to favor emotionally charged, sensational or polarizing content. Generative AI makes it possible to produce and distribute convincing synthetic material on an unprecedented scale, weakening traditional connections between information, authorship and authenticity[6]. Without a sufficiently shared and reliable knowledge environment, citizens cannot evaluate public issues effectively and democratic institutions lose their capacity for collective learning and self-correction.

Recommendation systems can shape visibility and incentives, but their effects differ across platforms, audiences, issues, and political settings.

Protecting this epistemic infrastructure requires pluralistic and independent journalism, public media, academic freedom, open science, libraries, education and fact-checking. Platforms should provide regulators and qualified researchers with sufficient access to examine recommendation systems and their social effects. Political advertising and microtargeting require transparency, restrictions on sensitive profiling and safeguards against covert foreign interference[7]. Digital and media literacy must become a lifelong democratic competence, enabling citizens to assess sources, recognize algorithmic influence and distinguish reliable information from manipulation or AI-generated content. These safeguards should protect an open and plural public sphere rather than establish a centralized authority over truth.

Data governance must likewise extend beyond individual consent[8] and conventional privacy protection. Data can generate public value in healthcare, science, education, mobility and environmental policy, but large-scale collection can also increase surveillance, discrimination, dependence and corporate power. Institutions such as data trusts and data commons can combine individual protections with socially beneficial use under transparent public conditions. Digital rights should therefore encompass privacy, control over personal data, access to essential digital services, protection against algorithmic arbitrariness, interoperability, inclusion and effective legal remedies.

Data trusts, data commons, and public data intermediaries should be presented as emerging institutional experiments rather than settled solutions. Their legitimacy depends on clear fiduciary duties, representation of data subjects and affected communities, security, purpose limitation, independent oversight, and exit or correction rights[9].

Essential digital infrastructure should be treated as a matter of public interest and democratic resilience. Secure digital identities, interoperable public registries, open communication standards, payment systems and shared data facilities may require public provision or strong public regulation. Open-source software and digital public goods increase transparency, security and independence from individual suppliers. Public or cooperative alternatives to commercial communication platforms can further increase institutional diversity. Such alternatives need not eliminate private enterprise; they provide counterweights and create a common foundation upon which private innovation can develop.

Dependence on a small number of global providers of cloud services, advanced chips, satellites and AI systems also creates geopolitical risks. Digital sovereignty should not mean technological isolation or complete national self-sufficiency. It means maintaining sufficient capacity to protect essential services and democratic choices through European cooperation, supplier diversification[10], shared cybersecurity, open standards and strategically governed infrastructure[11].

No single institution possesses enough expertise, legitimacy or authority to govern the digital domain alone. Parliaments must establish democratic objectives and legal limits; independent regulators must supervise compliance; courts must protect rights and provide remedies; European institutions must address cross-border markets; researchers must evaluate technologies independently; civil-society organizations must represent vulnerable interests; journalists must expose abuses; open-source communities must improve transparency; and citizens must contribute experience and democratic judgement. Their complementary roles create a polycentric system in which digital power remains open to correction[12].

For every high-impact digital system, responsibility should be allocated before deployment. The provider should document design and limitations; the deployer should justify the purpose and context of use; an independent authority should supervise compliance; qualified auditors and researchers should test systemic effects; courts or tribunals should provide remedies; and affected persons should receive notice, assistance, and a practical route to challenge. Contracting out technology must never mean contracting out public responsibility.

Because technology develops faster than legislation and its consequences often become visible only after widespread adoption, digital governance must remain adaptive. Regulatory experimentation, controlled testing environments, periodic evaluations, sunset clauses and mandatory reviews can support institutional learning while preserving legal safeguards. Conflicts between innovation and rights, transparency and commercial confidentiality, security and openness, or misinformation controls and freedom of expression should be addressed through proportionality, judicial review and accessible remedies.

A democratic digital order must ultimately be assessed not only by efficiency, growth or technological progress, but also by dignity, equality, autonomy, inclusion, resilience and democratic legitimacy. Digitalization can support human development only when its underlying systems remain transparent, contestable and institutionally correctable. The decisive safeguard is therefore a diverse architecture of countervailing powers in which no government, corporation or technical system can exercise lasting control without democratic and legal accountability.

Would you like to explore how a relational understanding of humanity can transform democracy, law, the economy, and public institutions? Click the link below to read the full book, Becoming Human Together, and discover a new framework for building a more humane, democratic, and correctable society within ecological limits.

https://www.researchgate.net/publication/413458836_Becoming_Human_Together_Rethinking_Democracy_Power_and_Institutions_for_a_Complex_and_Changing_World

---

 

[2] European Union, Regulation (EU) 2022/1925 on contestable and fair markets in the digital sector (Digital Markets Act), OJ L 265, 12 October 2022, https://eur-lex.europa.eu/eli/reg/2022/1925/oj/eng.

[3] Andrew D. Selbst et al., “Fairness and Abstraction in Sociotechnical Systems,” Proceedings of FAT* ’19 (2019): 59–68, https://doi.org/10.1145/3287560.3287598; Safiya Umoja Noble, Algorithms of Oppression (New York: New York University Press, 2018).

[4] European Union, Regulation (EU) 2024/1689 (Artificial Intelligence Act), https://eur-lex.europa.eu/eli/reg/2024/1689/oj/eng.

[5] Joshua A. Kroll et al., “Accountable Algorithms,” University of Pennsylvania Law Review 165, no. 3 (2017): 633–705; Article 22 of Regulation (EU) 2016/679 (General Data Protection Regulation), OJ L 119, 4 May 2016, https://eur-lex.europa.eu/eli/reg/2016/679/oj/eng.

[6] Emily M. Bender et al., “On the Dangers of Stochastic Parrots,” Proceedings of FAccT ’21 (2021): 610–623, https://doi.org/10.1145/3442188.3445922; UNESCO, Recommendation on the Ethics of Artificial Intelligence, 2021.

[7] Frederik J. Zuiderveen Borgesius et al., “Online Political Microtargeting,” Utrecht Law Review 14, no. 1 (2018): 82–96, https://doi.org/10.18352/ulr.420; Regulation (EU) 2024/900 on the transparency and targeting of political advertising, OJ L, 2024/900, 20 March 2024.

[8] European Union, Regulation (EU) 2022/868 on European data governance (Data Governance Act), OJ L 152, 3 June 2022, https://eur-lex.europa.eu/eli/reg/2022/868/oj/eng; Regulation (EU) 2023/2854 on harmonised rules on fair access to and use of data (Data Act), OJ L, 2023/2854, 22 December 2023.

[9] Sylvie Delacroix and Neil D. Lawrence, “Bottom-up Data Trusts,” International Data Privacy Law 9, no. 4 (2019): 236–252, https://doi.org/10.1093/idpl/ipz014. The legal form, fiduciary duties, and representation model remain unsettled.

[10] Farrell and Newman, “Weaponized Interdependence”; European Union, Directive (EU) 2022/2555 on measures for a high common level of cybersecurity across the Union (NIS 2 Directive), OJ L 333, 27 December 2022.

[11] United Nations Development Programme, The DPI Approach: A Playbook (New York: UNDP, 2023); European Commission, Open Source Software Strategy 2020–2023. Public provision does not by itself guarantee accountability or security.

[12] Ostrom, “Beyond Markets and States”; European Commission, 2026 Rule of Law Report. Explicitly connect digital oversight to courts, regulators, parliaments, researchers, civil society, and accessible remedies.