Quick Buzz Feed

Writing Artificial Intelligence Into Constitutions: A Global Constitutional Moment for the Algorithmic State

Gary Lloyd | Oct 21,26 | 01:34 EST

Politics

Boldizsár Szentgáli-Tóth, senior research fellow of the ELTE Centre for Social Sciences, Institute for Legal Studies (Budapest) and lawyer at the K-Monitor Public Benefit Association (Budapest), and Rebeka Kiss, junior research fellow of the ELTE Centre for Social Sciences, Institute for Legal Studies (Budapest), explore the constitutional implications of Artificial Intelligence.

The instinct to reach upward The drive to incorporate AI into constitutional frameworks stems from a profound intuition that fundamental laws offer a level of durability and symbolic authority that ordinary statutes cannot match. This phenomenon is often termed 'algorithmic constitutionalism,' recognizing that Artificial Intelligence is fundamentally reshaping how both public and private power operates, thereby necessitating a re-evaluation and reframing of existing fundamental rights. Examples like Chile's 2021 constitutional reform on neurorights, which protects brain activity and associated data, highlight a global trend where states perceive certain technological advancements as serious enough to warrant inclusion in their foundational documents. The core question for drafters is why algorithmic power, with its wide-ranging societal impact, should not receive comparable constitutional safeguards, mirroring the protections extended to neurotechnology. Why entrenchment is the wrong tool Constitutionalizing AI presents significant challenges due to the rapid evolution of the technology itself. AI is a dynamic collection of techniques, constantly changing in capabilities and terminology, which makes any rigid constitutional definition quickly obsolete. Constitutions are deliberately designed for stability across generations, making them unsuitable for such a fast-moving field. Furthermore, highly specific AI clauses risk undermining the adaptive power of abstract constitutional guarantees like human dignity or equal treatment. Courts have historically applied these broad rights to unforeseen circumstances, such as automated police data analysis, and narrowly drafted AI provisions could lead to an 'expressio unius' argument, limiting protection to only explicitly mentioned aspects. Lastly, entrenchment can create a false sense of security; a vague constitutional slogan on AI respecting human dignity is largely redundant and avoids the complex technical work (e.g., human oversight, explainability, liability) that belongs in more flexible statutory and regulatory instruments like the EU AI Act or GDPR. This argues for a cautious approach to direct constitutional entrenchment of AI. Courts extending old guarantees The existing legal architecture, particularly the principles of the 'Rechtsstaat' (rule of law), already provides robust tools for governing the algorithmic state without the need for novel constitutional provisions specific to AI. When public administration employs automated systems for individual decision-making, established principles such as legality, the duty to provide reasons, and the right to effective judicial protection automatically apply. International and national courts have demonstrated this adaptive capacity. For instance, a Dutch court, invoking Article 8 of the European Convention on Human Rights, halted the SyRI welfare-fraud system due to insufficient transparency and lack of notification. Similarly, France's Conseil constitutionnel, reviewing the Parcoursup admissions algorithm, mandated publication of institutional criteria based on an 18th-century accountability clause. These cases underscore a shared principle of reviewability: the algorithmic exercise of public power is not exempt from accountability and the right to remedy, proving that abstract legal guarantees can effectively address modern technological challenges without explicit AI mentions. A global patchwork Globally, states and international bodies are developing a diverse range of responses to govern AI within a constitutional or quasi-constitutional context. A significant development is the Council of Europe's Framework Convention on Artificial Intelligence, which, as the first binding international treaty on AI, enshrines principles of accountability, transparency, and non-discrimination for public authorities' use of AI, demonstrating that robust commitments can be made without specific technological nomenclature. At the national level, some jurisdictions are moving to explicitly name AI in their constitutions or related reforms. Greek Prime Minister Kyriakos Mitsotakis proposed incorporating AI into the Greek constitution to safeguard individual freedom and social welfare. Brazil adopted a constitutional amendment protecting personal data, sidestepping direct AI mention but addressing a closely related concern. Conversely, a 2026 amendment proposal in Minnesota, aiming to deny AI free-speech rights, illustrates the risks of poorly drafted clauses potentially infringing upon human rights exercised through AI tools. This highlights two main strategies: entrenching explicit digital rights/objectives or adapting existing rights through judicial interpretation, with the latter often proving more flexible and resilient. The narrow residue While the bulk of AI governance should be flexible and reside in legislation, a 'narrow residue' of three abstract, durable principles warrants constitutional entrenchment, primarily binding public power and influencing private power through horizontal effects. First, the **primacy of human judgment** ensures that any public decision significantly affecting an individual requires genuine, accountable human deliberation, prohibiting automatic sanctions or service exclusions. This principle safeguards democratic accountability and the non-delegability of core state functions, making it suitable for a constitutional text. Second, **equitable access to AI as a state objective**. Recognizing AI's increasing role, constitutions can establish this as a guiding value, obliging the state to ensure fair access and combat digital exclusion, thereby supporting equality of opportunity and strengthening public accountability. Third, **transparency of public-sector AI as a principle**. This mandates public knowledge of the existence, funding, developers, purpose, and operating logic of public-sector decision-support systems, aligning with freedom of information and accountability. Initiatives like public algorithm registers (e.g., Amsterdam, Helsinki, Netherlands) exemplify how this principle can be translated into practical implementation, demonstrating enduring constitutional commitments that transcend specific technologies. Conclusion The debate surrounding the constitutionalization of Artificial Intelligence fundamentally centers on identifying the appropriate level of legal abstraction. Highly detailed and easily revisable rules for AI governance are best placed in legislation and international instruments, allowing them to adapt dynamically to technological advancements. In contrast, foundational commitments concerning individual rights and the exercise of public power are largely already enshrined in existing constitutional frameworks and can be effectively applied through judicial interpretation, rather than necessitating wholesale rewriting. The authors caution against the potential pitfall of embedding fashionable but transient AI definitions into constitutions, which could lead to superficial governance and quickly become obsolete. Comparative experience demonstrates that existing fundamental rights can be successfully adapted to govern the algorithmic state, provided that there is a strong political will and institutional capacity for enforcement. Therefore, during the drafting of new constitutional texts, the critical task is to precisely identify a very narrow set of enduring, abstract principles that genuinely cannot be addressed by existing guarantees, rather than pursuing broad, potentially restrictive, or soon-to-be-outdated AI-specific provisions.

The instinct to reach upward

The drive to incorporate AI into constitutional frameworks stems from a profound intuition that fundamental laws offer a level of durability and symbolic authority that ordinary statutes cannot match. This phenomenon is often termed 'algorithmic constitutionalism,' recognizing that Artificial Intelligence is fundamentally reshaping how both public and private power operates, thereby necessitating a re-evaluation and reframing of existing fundamental rights. Examples like Chile's 2021 constitutional reform on neurorights, which protects brain activity and associated data, highlight a global trend where states perceive certain technological advancements as serious enough to warrant inclusion in their foundational documents. The core question for drafters is why algorithmic power, with its wide-ranging societal impact, should not receive comparable constitutional safeguards, mirroring the protections extended to neurotechnology.

Why entrenchment is the wrong tool

Constitutionalizing AI presents significant challenges due to the rapid evolution of the technology itself. AI is a dynamic collection of techniques, constantly changing in capabilities and terminology, which makes any rigid constitutional definition quickly obsolete. Constitutions are deliberately designed for stability across generations, making them unsuitable for such a fast-moving field. Furthermore, highly specific AI clauses risk undermining the adaptive power of abstract constitutional guarantees like human dignity or equal treatment. Courts have historically applied these broad rights to unforeseen circumstances, such as automated police data analysis, and narrowly drafted AI provisions could lead to an 'expressio unius' argument, limiting protection to only explicitly mentioned aspects. Lastly, entrenchment can create a false sense of security; a vague constitutional slogan on AI respecting human dignity is largely redundant and avoids the complex technical work (e.g., human oversight, explainability, liability) that belongs in more flexible statutory and regulatory instruments like the EU AI Act or GDPR. This argues for a cautious approach to direct constitutional entrenchment of AI.

Courts extending old guarantees

The existing legal architecture, particularly the principles of the 'Rechtsstaat' (rule of law), already provides robust tools for governing the algorithmic state without the need for novel constitutional provisions specific to AI. When public administration employs automated systems for individual decision-making, established principles such as legality, the duty to provide reasons, and the right to effective judicial protection automatically apply. International and national courts have demonstrated this adaptive capacity. For instance, a Dutch court, invoking Article 8 of the European Convention on Human Rights, halted the SyRI welfare-fraud system due to insufficient transparency and lack of notification. Similarly, France's Conseil constitutionnel, reviewing the Parcoursup admissions algorithm, mandated publication of institutional criteria based on an 18th-century accountability clause. These cases underscore a shared principle of reviewability: the algorithmic exercise of public power is not exempt from accountability and the right to remedy, proving that abstract legal guarantees can effectively address modern technological challenges without explicit AI mentions.

A global patchwork

Globally, states and international bodies are developing a diverse range of responses to govern AI within a constitutional or quasi-constitutional context. A significant development is the Council of Europe's Framework Convention on Artificial Intelligence, which, as the first binding international treaty on AI, enshrines principles of accountability, transparency, and non-discrimination for public authorities' use of AI, demonstrating that robust commitments can be made without specific technological nomenclature. At the national level, some jurisdictions are moving to explicitly name AI in their constitutions or related reforms. Greek Prime Minister Kyriakos Mitsotakis proposed incorporating AI into the Greek constitution to safeguard individual freedom and social welfare. Brazil adopted a constitutional amendment protecting personal data, sidestepping direct AI mention but addressing a closely related concern. Conversely, a 2026 amendment proposal in Minnesota, aiming to deny AI free-speech rights, illustrates the risks of poorly drafted clauses potentially infringing upon human rights exercised through AI tools. This highlights two main strategies: entrenching explicit digital rights/objectives or adapting existing rights through judicial interpretation, with the latter often proving more flexible and resilient.

The narrow residue

While the bulk of AI governance should be flexible and reside in legislation, a 'narrow residue' of three abstract, durable principles warrants constitutional entrenchment, primarily binding public power and influencing private power through horizontal effects. First, the **primacy of human judgment** ensures that any public decision significantly affecting an individual requires genuine, accountable human deliberation, prohibiting automatic sanctions or service exclusions. This principle safeguards democratic accountability and the non-delegability of core state functions, making it suitable for a constitutional text. Second, **equitable access to AI as a state objective**. Recognizing AI's increasing role, constitutions can establish this as a guiding value, obliging the state to ensure fair access and combat digital exclusion, thereby supporting equality of opportunity and strengthening public accountability. Third, **transparency of public-sector AI as a principle**. This mandates public knowledge of the existence, funding, developers, purpose, and operating logic of public-sector decision-support systems, aligning with freedom of information and accountability. Initiatives like public algorithm registers (e.g., Amsterdam, Helsinki, Netherlands) exemplify how this principle can be translated into practical implementation, demonstrating enduring constitutional commitments that transcend specific technologies.

Conclusion

The debate surrounding the constitutionalization of Artificial Intelligence fundamentally centers on identifying the appropriate level of legal abstraction. Highly detailed and easily revisable rules for AI governance are best placed in legislation and international instruments, allowing them to adapt dynamically to technological advancements. In contrast, foundational commitments concerning individual rights and the exercise of public power are largely already enshrined in existing constitutional frameworks and can be effectively applied through judicial interpretation, rather than necessitating wholesale rewriting. The authors caution against the potential pitfall of embedding fashionable but transient AI definitions into constitutions, which could lead to superficial governance and quickly become obsolete. Comparative experience demonstrates that existing fundamental rights can be successfully adapted to govern the algorithmic state, provided that there is a strong political will and institutional capacity for enforcement. Therefore, during the drafting of new constitutional texts, the critical task is to precisely identify a very narrow set of enduring, abstract principles that genuinely cannot be addressed by existing guarantees, rather than pursuing broad, potentially restrictive, or soon-to-be-outdated AI-specific provisions.