Beyond Consent: Algorithmic Welfare, Constitutive Dependence, and the Limits of Liberal AI Ethics

The paradigm is Aadhaar, India's biometric identity infrastructure, on which food rations, pensions, school meals and wages now depend. The elderly labourer whose worn fingerprints fail at the ration shop is not failing to give informed consent; she is failing to be recognised as someone who could give or withhold it. Aadhaar does not act on a standing she already has but produces her operative standing as an entitlement-bearer, and she cannot refuse it without refusing the standing from which refusal would make sense. I call this condition constitutive dependence.

The constructive half of the paper asks what such an institution owes in place of consent. Drawing on the recognition-theoretic tradition of Honneth, Fraser and Butler, it states the Constitutive Obligation Principle: where an institution produces its subjects' operative standing, its legitimacy depends not on their authorisation but on discharging non-waivable obligations of legibility, non-abandonment and remediation.

The paper reads Aadhaar not as an outlier but as an early instance of an architectural form towards which algorithmic governance is converging, and offers the principle as a framework for the algorithmic welfare state more broadly.

The main claims

1. In data-protection-derived regimes and in the remedial debate around algorithmic welfare, consent carries the load of legitimation and redress.
2. Even the most sophisticated critics of consent retain its presupposition: a subject who exists prior to, and independently of, the system.
3. Algorithmic welfare systems such as Aadhaar can produce the operative standing of the people they govern: the condition of constitutive dependence.
4. Under constitutive dependence consent fails conceptually, and not only practically, because no standpoint of refusal preserves the standing at stake.
5. Legitimacy then rests on the Constitutive Obligation Principle: non-waivable obligations of legibility, non-abandonment and remediation.

Abstract

Within the data-protection-derived legal and governance frameworks that regulate algorithmic systems (the GDPR, the EU AI Act, and the principlist codes that track them), consent and individual authorisation are the load-bearing mechanism of legitimation and redress, and the same frame organises the remedial debate over large-scale algorithmic welfare systems, the Indian Aadhaar architecture being the paradigm case. This paper argues that for a distinctive class of such systems consent is not merely practically strained but conceptually misplaced, and develops a positive alternative. Its central contribution is the concept of constitutive dependence: a condition in which an algorithmic system does not constrain or coerce agents recognised as rights-bearers but produces their operative standing as subjects of recognition and entitlement. Where it obtains, the standpoint of refusal on which consent depends is foreclosed by the system’s algorithmic architecture, which enforces its criteria of recognition without the discretionary judgement by which a human-mediated bureaucracy can recognise the manifestly entitled.

Locating the condition within the recognition-theoretic tradition (Honneth, Fraser, Butler; Waelen) as a distinctively infrastructural form of misrecognition, the paper articulates the Constitutive Obligation Principle: where an institution produces its subjects’ operative standing as entitlement-bearers, its legitimacy depends not on their authorisation but on discharging non-waivable obligations of legibility, non-abandonment, and remediation. Developed through the Aadhaar case and offered as a framework for the algorithmic welfare state more broadly, the argument reads Aadhaar not as an outlier but as an early instance of an architectural form toward which algorithmic governance is converging.

@article{saxena2026beyond,
  author  = {Saxena, Abhinav},
  title   = {Beyond Consent: Algorithmic Welfare, Constitutive Dependence, and the Limits of Liberal AI Ethics},
  journal = {AI and Ethics},
  year    = {2026},
  volume  = {6},
  pages   = {414},
  doi     = {10.1007/s43681-026-01270-5},
}