Artificial intelligence is usually discussed through questions about what AI systems can do. Can they predict behaviour? Rank applicants? Generate content? Assess risk? Recommend decisions?
These questions matter, but they can obscure another dimension of AI power: who controls the infrastructures through which AI operates, and who defines the rules governing participation within them?
In The Inequality Regime of AI, Massimo Ragnedda and Maria Laura Ruiu address this dimension through two connected concepts: Digital Feudalism and Lex Digitalis.
Digital Feudalism describes forms of dependency that emerge when individuals, organisations, institutions, developers, businesses, and public authorities increasingly rely on digital infrastructures controlled by powerful private actors.
Lex Digitalis refers to the private rules embedded in code, algorithms, terms of service, ranking systems, platform policies, technical architectures, and automated enforcement mechanisms.
The distinction between the two concepts is important. Digital Feudalism identifies a structure of dependency. Lex Digitalis identifies the architecture of private rule operating within that structure.
Together, they help explain a dimension of AI inequality that cannot be reduced to access, digital skills, or algorithmic bias.
What is Digital Feudalism?
Contemporary social and economic life increasingly depends on privately controlled digital infrastructures.
Cloud computing, AI models, application programming interfaces, app stores, platforms, data infrastructures, search systems, and communication networks have become central to the activities of individuals, businesses, developers, universities, governments, and other institutions.
Using a privately owned service does not automatically constitute Digital Feudalism. The concept becomes analytically useful when dependence on an infrastructure produces highly asymmetrical relationships of power.
Consider a developer whose application depends on access to an AI model owned by another company. A business may depend on a platform to reach its customers. A creator may rely on a particular digital environment for visibility and income. A public institution may integrate privately developed AI systems into services used by citizens.
In these cases, participation occurs through infrastructures that users may neither own nor meaningfully govern.
The organisation controlling the infrastructure may change access conditions, modify technical standards, alter pricing, redesign interfaces, restrict functionalities, or remove access entirely.
Digital Feudalism therefore directs attention not simply to ownership, but to dependency, bargaining power, control of essential infrastructures, and the possibility of exit.
The term should not be interpreted as suggesting that contemporary digital societies reproduce historical feudalism. Rather, the analogy highlights relationships in which participation increasingly depends on resources controlled by actors with far greater power to determine the conditions under which those resources may be used.
This raises a basic question:
What happens when participation in social, economic, institutional, and cultural life increasingly depends on infrastructures that users cannot meaningfully control?
What is Lex Digitalis?
Dependency on infrastructure is only one part of the problem.
Digital environments also contain rules.
Some are visible, such as terms of service or platform policies. Others are embedded within the technical architecture itself.
Code determines what actions are possible.
Algorithms influence what becomes visible.
Ranking systems establish priorities.
Recommendation systems organise attention.
Moderation systems determine what may remain accessible.
Access protocols establish who can use particular services and under what conditions.
Automated enforcement can restrict, suspend, downgrade, exclude, or prioritise participants.
We use the concept of Lex Digitalis to capture this increasingly important form of private rule-making.
The term draws attention to how governance in digital environments does not operate only through legislation, courts, and public institutions. Rules can also be produced and enforced through technical systems designed by private organisations.
Users may therefore experience governance without encountering anything that looks like conventional law.
A change to an algorithm can transform visibility.
A change to an API can affect an entire business model.
A modification to an AI system can alter which forms of interaction are permitted.
A ranking system can reorganise opportunities without issuing anything resembling a legal decision.
Lex Digitalis therefore raises another fundamental question:
Who makes the rules of digital participation, and through what mechanisms are those rules enforced?
When infrastructure becomes governance
Digital Feudalism and Lex Digitalis become particularly significant when considered together.
If people and institutions depend on privately controlled digital infrastructures, those controlling the infrastructure may also acquire considerable power to determine the rules operating within it.
Control over infrastructure can therefore become rule-making power.
This relationship is especially important in AI.
AI systems increasingly depend on concentrated infrastructures involving cloud computing, computing capacity, proprietary models, datasets, interfaces, APIs, and platform ecosystems.
The question is therefore not only who develops AI systems or how accurate those systems are.
It is also:
Who owns the infrastructures?
Who controls access to advanced AI models?
Who determines the conditions under which developers and institutions can use them?
Who can modify those conditions?
Who can challenge those decisions?
And how realistic is meaningful exit when an infrastructure becomes central to economic or institutional participation?
These questions connect AI governance directly to questions of power and inequality.
Digital Feudalism, Lex Digitalis, and the AI Inequality Regime
Digital Feudalism and Lex Digitalis form part of the wider framework developed in The Inequality Regime of AI.
The book argues that AI inequality cannot be understood only through unequal access to technology or through biased outputs produced by individual algorithms. AI should instead be analysed as part of a broader social, institutional, and infrastructural system through which visibility, value, voice, opportunities, resources, and risk are organised.
Other concepts developed within this framework examine different dimensions of this process.
The AI Stratification Spiral focuses on how historical inequalities may become embedded in data, predictions, institutional decisions, and subsequent data generation.
The Allocative Turn shifts attention from who can access digital technologies towards who controls systems that classify, rank, and allocate opportunities.
Algorithmic Habitus examines how repeated exposure to algorithmically organised environments may shape expectations and behaviour.
Intelligibility Inequality concerns unequal capacities to be accurately recognised and represented by computational systems.
Digital Feudalism and Lex Digitalis add the infrastructural and governance dimensions.
They ask who owns the environments within which these processes occur and who possesses the authority to establish their operating rules.
This is why understanding AI inequality requires examining more than algorithms.
It requires examining the infrastructures that make AI possible, the dependencies these infrastructures create, and the systems of private rule through which access and participation are organised.
The central issue can therefore be expressed simply:
Digital Feudalism concerns who controls the infrastructure. Lex Digitalis concerns who writes the rules.
Together, they invite us to ask a broader question about the emerging organisation of AI societies:
When participation increasingly depends on privately controlled digital infrastructures, who should have the power to determine the conditions under which everyone else participates?