It is not every day that a major AI company invokes constitutional values against the US government. Anthropic – the US-American AI company behind Claude – declined the US military’s request for unrestricted access to its AI tools, citing worries about domestic mass surveillance and the use of its technology in fully autonomous weapons. As a result, a standoff developed between Anthropic and the US government, and the company was labelled a “supply chain risk.” This means businesses working with the military can no longer do business with Anthropic.

Recognising digital corporations as potential drivers of constitutional rights runs counter to the most recent literature on digital regulation and digital sovereignty. However, perceiving them as such remains crucial in a time when constitutional values are questioned or disregarded even within traditionally established constitutional democracies.

Platform governance matrix

To better understand the interplay among the different actors in digital regulation, I suggest a governance-matrix perspective that views digital governance as a multi-level system of competing authority and legitimacy claims among global institutions, nation-states, digital platforms, and individuals. While the UN and regional blocs (e.g., the EU) frame supranational norms that shape state authority, policies, and economic commitments, the nation-state occupies a pivotal mediating position, asserting sovereignty over platforms while deriving democratic legitimacy (or the appearance thereof) from citizens. Digital platforms, in turn, oscillate between compliance and defiance vis-à-vis states’ regulatory efforts, either for profit or to assess their role in advancing their users’ well-being in politically contested environments. At the same time, platforms exercise power over users through contractual frameworks built on notions of trust and consent, which, given their largely unilateral design, risk concentrating authority in ways that are difficult to contest. As Balkin puts it, ‘private governors are much like nineteenth century enlightened despots’, implementing values their end-users might want while reserving the right to act arbitrarily.

The matrix also reveals a basic instability at the heart of platform governance: authority is dispersed across multiple levels and constantly negotiated. Global institutions articulate supranational norms, states invoke sovereignty and democratic legitimacy, platforms control infrastructures that shape communication and access, and users are both protected and constrained by rules they do not meaningfully shape. None of these actors exercises uncontested authority. Seen this way, governance in the digital sphere is less a settled legal order than an ongoing struggle over who gets to set norms, enforce them, and justify them. This matters because calls for stronger state control over tech firms can look very different across political contexts and over time. While such measures may reinforce rights protections in consolidated democracies, they may enable further restrictions where democratic institutions are weakened or authoritarian tendencies prevail.

The legal and constitutional dimensions of the feud between Anthropic and the US government are clear when one considers the division of labour between private companies and public authorities in protecting constitutional values, fundamental rights, and individual autonomy. Anthropic’s decision not to meet the US Department of Defense demands fits a pattern seen in other parts of the world. For instance, in 2021, Twitter refused to suspend accounts and remove content about protests in India, saying those requests might violate constitutional and international human rights standards. After Anthropic was labelled a supply chain risk, CEO Dario Amodei explained that the company views mass surveillance and autonomous weapons as red lines that should not be crossed without proper oversight and regulation. He also said the company is “patriotic” and believes that “crossing those red lines is contrary to American values”.

This ambivalence has long been visible in debates about digital regulation and governance. On one hand, as early as 2001, Sunstein diagnosed the risks of digital communication to deliberative democracy, focusing primarily on polarization, echo chambers, and information cocoons, and argued for a stronger role for the state in structuring the digital public sphere. On the other hand, Chander warned that, beyond these risks, digital communications remain an important tool for information dissemination and the “project of deepening democracy to include marginalised groups”, suggesting caution when arguing for purely state regulatory control. More recently, engaging with literature on the Brussels Effect and the DSA, Chander also raised awareness of the possibility that the regulation could create “digital czars” if its provisions are instrumentalised by authoritarian leaders to expand their control over speech rights.

Concurring narratives and EU digital constitutionalism

The discomfort with regulatory frameworks that prioritise public powers might stem from legal and cultural distinctions between US and European scholars and policymakers. In the United States, strong First Amendment traditions and a deep-seated suspicion of state interference in speech often frame market-freedoms and private ordering as preferable to federal government control. In Europe, by contrast, constitutional practice is more accustomed to balancing fundamental rights through proportionality analysis and to embedding such protections within legislative frameworks such as the DSA. Moreover, while American debates frequently focus on limiting government overreach, European regulatory culture has been more willing to view concentrated corporate power as a constitutional concern in its own right. However, the increasing number of constitutional crises may be relevant to future assessments of regulatory approaches on both sides of the pond.

In Europe, digital constitutionalism promotes a policy approach towards digital governance focused on preventing private interests from unduly influencing constitutional law. It seeks to restore balance between public and private authorities, protect constitutional democracies from being sidelined, and respond to the negative impacts of digitalization, especially the rise of a “power-profit complex” that fuels digital totalitarianism and hinders positive digital development. This perspective highlights concerns that private companies prioritize profit over the public good and urges state and regional bodies, such as the EU, to establish robust regulatory frameworks to limit corporate influence. Echoing Bradford’s view of digital regulation as a “vertical war”, it rejects the idea that platforms can effectively self-regulate in society’s interest, instead emphasizing the role of governments in ensuring that human rights are upheld in the digital sphere.

This constitutionalist mindset, focused on reframing the balance between public and private interests, is clear in the DSA’s overall regulatory framework, which accepts that online platforms maintain powers and responsibilities, while commandeering their governance mandate by establishing a set of rights, obligations, and a risk assessment framework that prioritizes public interests (such as fundamental rights) over profit concerns and design choices. It does so by adopting a horizontal approach that integrates and operationalises fundamental rights protections in EU law and sector-specific legislation, while establishing a mixed enforcement system that combines, among others, private mechanisms, national Digital Services Coordinators, and the European Commission.

However, digital constitutionalism’s aspirations cannot be confined solely to efforts to promote democracy and fundamental rights through national or international regulatory frameworks, assuming that the nation-state will always promote the well-being of its citizens. Purely public regulation, focused on individual rights, courts, and due process, is not enough when liberal democracies are further encroached by a rule-of-law crisis, diminishing commitments to constitutional democracy, and the expansion of authoritarian and neo-fascist discourses.

Anthropic’s current standoff in the United States indicates that the issue is not necessarily when a private and despotic decision is made in favour of citizens’ rights and constitutional democratic values, even if it may be surprising in stable democracies where citizens have strong representative and participatory rights. These constitutional decisions are highly political, and without a structured accountability framework centred on citizens’ rights and interests, they risk foregrounding business opportunities over individual and collective autonomy interests. This ambivalence has already been made clear by the general alignment of big tech companies with the US government following the start of Trump’s second mandate, including their support for the US government’s intensified antagonism toward the EU’s regulatory efforts.

Concerns for a digitally mediated constitutional landscape in Europe

While from a global perspective, Chander’s assessment of possible digital czars is relevant, within the EU itself, the multilevel framework of the DSA may help prevent this dynamic from materialising. The harmonisation of due diligence and intermediary liability rules across member states through EU law de facto prevents national authorities from usurping powers to expand their control over speech rights. Even if such measures are adopted and supported at the constitutional level, the Commission and the European Court of Justice would maintain jurisdiction over those attempts through infringement proceedings, however long and politically costly that may be.

Still, attention towards amplifying the legitimacy of private decisions in the digital environment beyond state regulatory mechanisms remains relevant. A strong commitment to constitutional values underpins the effectiveness of regulatory efforts to curb the dark patterns in the digital ecosystem. The Commission’s framing of its infringement proceedings against Hungary (C-769/2022) showcases how EU fundamental values are not a monolith and require further interpretation and normative application. Moreover, while American companies may be willing to protect American values and interests, recently, some of these companies have also aligned politically with right-wing and neo-fascist movements within Europe, consolidating a much more dynamic political repercussion of their services.

Economic exit decisions might serve as one mechanism for further influencing big tech companies to abide by constitutional and democratic commitments. The current “QuitGPT” campaign – following OpenAI’s overtake of Anthropic’s contract with the US government – is one example. However, this strategy is effective only to the extent that there are alternative services with lower transition costs. In Europe, that leverage is limited by a lack of native alternatives.

Digital constitutionalism, therefore, should not focus only on state regulation. It must also be considered as a guide towards enhancing direct political contestation of big techs and their controlling shareholders’ broader geopolitical performance. As the Anthropic episode clarifies, constitutional tensions may also arise when private actors resist state demands in the name of fundamental values. The challenge, then, is not simply to shift power from platforms to states, but to construct a digitally mediated constitutional landscape in which private and public powers are both subject to structured accountability – and in which individuals remain the ultimate beneficiaries of constitutional guarantees.