Explanatory note

This model Charter proposes a universal, technology-neutral framework for governing artificial intelligence throughout its lifecycle. It is designed for consideration by States, international organisations, civil society, academia, workers, affected communities and the private sector. Its central proposition is that artificial intelligence shall serve human dignity, peace, human rights, sustainable development and the common good.

The text uses the form and vocabulary of multilateral instruments: a preamble, common purposes and principles, substantive obligations, institutions, implementation arrangements and final clauses. It combines binding-style minimum duties with national flexibility, proportional regulation and enhanced assistance for developing countries. Adoption would require an intergovernmental negotiating process and any legal review required by the competent organs of the United Nations.

The Charter is informed by the Charter of the United Nations and the international human rights framework; the UNESCO Recommendation on the Ethics of Artificial Intelligence; General Assembly resolution 78/265 on safe, secure and trustworthy artificial intelligence systems; the Global Digital Compact; and the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law. It does not displace stronger protections in international or domestic law.

Design of the proposed regime

Element Function Universal floor Minimum duties apply across public and private uses, without preventing stronger national protection.

Risk proportionality Obligations increase with the scale, likelihood, reversibility and distribution of harm.

Rights and remedy Affected persons receive notice, explanation where material, human review, contestability and effective remedy.

Global equity Capacity building, accessible technology, fair participation and benefit sharing address digital and development divides.

Adaptive governance Independent science, incident reporting, audits, periodic review and interoperable standards respond to technological change.

Preamble

The States Parties to the present Charter, Resolved to save present and succeeding generations from the harms that may arise from the unsafe, inequitable or unlawful development and use of artificial intelligence, while enabling its benefits to advance peace, human welfare and sustainable development, Reaffirming faith in the purposes and principles of the Charter of the United Nations, in the dignity and worth of the human person, in the equal rights of all persons and of nations large and small, and in the obligations arising from international law, Recalling the Universal Declaration of Human Rights and the core international human rights treaties, and recognising that the same rights that people possess offline must be respected, protected and fulfilled where artificial intelligence is developed or used, Recognizing that artificial intelligence may contribute to education, health, science, culture, public administration, climate action, accessibility and the achievement of the Sustainable Development Goals, Mindful that artificial intelligence may also intensify discrimination, surveillance, manipulation, exclusion, labour displacement, environmental degradation, concentration of power, threats to information integrity and risks to international peace and security, Affirming that human beings and communities shall remain the subjects and beneficiaries of technological progress and shall not be reduced to objects of automated prediction, classification or control, Recognizing the sovereign equality of States and their differing social, cultural, legal and economic conditions, while affirming that sovereignty may not be invoked to derogate from applicable obligations under international law, Determined to close digital, data, computational, linguistic and knowledge divides and to secure the full and effective participation of developing countries, women and girls, children, persons with disabilities, Indigenous Peoples and other historically marginalized communities, Recognising the need for inclusive, transparent and evidence-based international cooperation involving States, international organisations, the private sector, technical communities, workers, academia, civil society and affected persons, Gravely concerned that the uncontrolled development of artificial intelligence exceeding general human capabilities could, through misalignment, autonomous replication, strategic deception or loss of effective human control, cause catastrophic or existential harm to humanity, Convinced that common principles, effective oversight and mutual assistance can support innovation worthy of public trust, Have agreed as follows:

---

Chapter I Purposes scope and definitions

Article 1 Purposes

1. The purposes of the present Charter are to ensure that artificial intelligence is developed and used for the benefit of humanity, in conformity with international law and with full respect for human dignity, human rights, democracy, the rule of law, peace and sustainable development.

2. The Charter establishes common minimum standards for the prevention and mitigation of harm, accountability throughout the artificial intelligence lifecycle, effective remedies, equitable access to benefits and international cooperation.

3. Nothing in the present Charter shall be interpreted as limiting a protection afforded by international law or by the law of a State Party where that protection is more conducive to human rights, safety or the public interest.

Article 2 Scope of application

1. The present Charter applies to activities within the lifecycle of artificial intelligence systems undertaken by public authorities and by natural or legal persons within the jurisdiction or effective control of a State Party.

2. Each State Party shall regulate conduct producing substantial effects within its jurisdiction, consistently with international law and principles of comity and cooperation.

3. The Charter applies without regard to the technique, model architecture, sector or form of deployment. Obligations shall be proportionate to risk and shall not be avoided by outsourcing, contractual allocation or technical design.

Article 3 Definitions

1. Artificial intelligence system means a machine-based system that, for explicit or implicit objectives, infers from the input it receives how to generate outputs, including predictions, content, recommendations or decisions, that can influence physical or virtual environments.

2. Lifecycle means research, design, data collection, development, testing, training, procurement, placing on the market, deployment, operation, monitoring, modification, retirement and decommissioning.

3. High-impact artificial intelligence system means a system whose use may materially affect human rights, safety, access to essential services, livelihoods, democratic processes, the environment or international peace and security.

4. General-purpose artificial intelligence model means a model capable of performing or being adapted to perform a broad range of tasks, including where incorporated into multiple downstream systems.

5. Advanced artificial intelligence system means a general-purpose or other system whose capabilities, autonomy, access to resources or capacity for adaptation could enable it materially to exceed human performance across a wide range of consequential tasks or to pose catastrophic or existential risk.

6. Affected person means any natural person or community whose rights, interests, opportunities, safety or environment are materially affected or reasonably likely to be affected by an artificial intelligence system.

7. Meaningful human control means timely, informed and effective human authority to understand the relevant context, supervise operation, intervene, refuse, override or terminate a system and assume responsibility for the resulting decision.

---

Chapter II Fundamental principles

Article 4 Human dignity and human agency

1. Human dignity, autonomy, bodily and mental integrity, and the capacity of persons to make informed choices shall guide every stage of the artificial intelligence lifecycle.

2. No person shall be subjected to an artificial intelligence system that negates legal personality, removes meaningful human responsibility or treats a person solely as an object of scoring, prediction or manipulation.

3. States Parties shall preserve meaningful human control wherever an artificial intelligence system may materially affect rights, safety or access to essential opportunities or services.

Article 5 Human rights equality and non discrimination

1. States Parties shall respect, protect and fulfil all internationally recognised human rights in relation to artificial intelligence systems and shall prevent direct, indirect, intersectional and systemic discrimination.

2. Design, data, testing and deployment shall take account of differences in sex, gender, age, race, colour, language, religion, political or other opinion, national or social origin, property, disability, birth, migration status, Indigenous identity and other status protected by international law.

3. Reasonable accommodation, accessibility and universal design shall be integral to artificial intelligence governance. Measures to achieve substantive equality shall not be considered discrimination.

Article 6 Safety security and resilience

1. Artificial intelligence systems shall be demonstrably safe, secure, robust and resilient throughout their lifecycle, having regard to foreseeable use, reasonably foreseeable misuse, human error and changing conditions.

2. States Parties shall require appropriate testing, evaluation, red teaming, cybersecurity, fail-safe mechanisms, incident response and post-deployment monitoring before and during high-impact use.

3. Uncertainty concerning serious or irreversible harm shall not be used as a reason to postpone proportionate preventive measures.

Article 7 Transparency explainability and traceability

1. Persons shall be informed when they are interacting with an artificial intelligence system or when artificial intelligence materially contributes to a decision affecting them, except where a lawful and strictly necessary limitation applies.

2. Information shall be understandable, accessible and proportionate to the context, including the system purpose, responsible entity, principal capabilities and limitations, material data sources, oversight arrangements and means of contesting an outcome.

3. States Parties shall require records sufficient to trace significant decisions, identify responsible actors and enable audit, investigation and remedy, with due protection for privacy, security and legitimate confidential information.

Article 8 Accountability and responsibility

1. Legal and natural persons shall remain accountable for decisions and actions taken with or through artificial intelligence. An artificial intelligence system shall not possess legal personality or bear responsibility in place of a person or institution.

2. States Parties shall allocate duties clearly among providers, deployers, importers, distributors, operators and public authorities, according to their knowledge, control and contribution to risk.

3. No contractual term shall exclude or diminish an obligation owed under the present Charter to an affected person or competent authority.

Article 9 Privacy and data governance

1. Personal data shall be processed lawfully, fairly, transparently and only for specified legitimate purposes, subject to necessity, proportionality, data minimisation, accuracy, security and appropriate retention limits.

2. States Parties shall safeguard privacy, confidentiality and freedom from arbitrary or unlawful surveillance, including through independent authorisation and oversight where required by international human rights law.

3. Data governance shall address provenance, consent where applicable, representativeness, quality, bias, intellectual property, community and collective interests, secure access and responsible deletion.

Article 10 Sustainability and shared benefit

1. Artificial intelligence shall be developed and used in a manner consistent with the right to development, intergenerational equity and the protection of the natural environment.

2. States Parties shall promote energy, water and material efficiency; measure and disclose significant environmental impacts; and prevent the externalization of environmental and social costs to vulnerable communities or future generations.

3. The benefits of artificial intelligence, including advances in science, education, health and accessibility, shall be shared fairly within and among countries.

---

Chapter III Duties of States Parties

Article 11 National implementation

1. Each State Party shall adopt or maintain legislative, administrative, judicial, budgetary and other measures necessary to give effect to the present Charter.

2. Each State Party shall designate one or more functionally independent competent authorities with adequate powers, expertise and resources to supervise compliance, investigate risks, order corrective action and cooperate internationally.

3. National frameworks shall provide coherent responsibilities across sectors, protect regulatory independence and include meaningful participation by affected persons and communities.

Article 12 Risk classification and assessment

1. States Parties shall establish a context-sensitive classification of artificial intelligence risks, taking account of severity, likelihood, scale, duration, reversibility, cumulative effects, vulnerability of affected persons and the possibility of misuse.

2. Before a high-impact system is placed in service, the responsible entity shall conduct and publish an accessible summary of an independent or independently reviewable impact assessment addressing human rights, equality, safety, privacy, labour, children, accessibility and environmental effects.

3. Assessment shall be continuous. Material modification, significant incidents or evidence of changed risk shall trigger reassessment, corrective action and, where necessary, suspension or withdrawal.

Article 13 Prohibited practices

1. States Parties shall prohibit an artificial intelligence practice where its nature or use is incompatible with human dignity, international human rights law or the principles of the present Charter and the risk cannot be adequately mitigated.

2. Prohibited practices shall include: social scoring by public authorities leading to unjustified or disproportionate treatment; intentional manipulation exploiting vulnerability in a manner likely to cause significant harm; real-time remote biometric identification in publicly accessible spaces except under law in strictly necessary and proportionate circumstances subject to independent authorisation; and inference of highly sensitive traits where not lawful, necessary and proportionate.

3. States Parties shall prohibit the design or deployment of an artificial intelligence system for torture, enforced disappearance, persecution, unlawful discrimination, mass arbitrary surveillance or the suppression of lawful democratic participation.

4. No artificial intelligence system shall make an irreversible decision to use nuclear weapons, nor shall any weapon system select and engage a human target without meaningful human control and compliance with international humanitarian law.

Article 14 Public authority use and procurement

1. Public authorities shall use artificial intelligence only under a clear legal mandate, for a legitimate purpose and where necessity, proportionality, competence and public benefit have been demonstrated.

2. No person shall be denied liberty, legal status, social protection, health care, education or another essential public service solely on the basis of an automated decision without effective human review.

3. Public procurement shall require lifecycle risk management, accessibility, audit access, data and security safeguards, incident disclosure, continuity arrangements, remedy and enforceable termination rights.

4. States Parties shall maintain a public register of high-impact artificial intelligence systems used by public authorities, subject only to lawful and narrowly tailored exceptions.

Article 15 Justice law enforcement and migration

1. The use of artificial intelligence in justice, law enforcement, border management, asylum and migration shall be subject to heightened legality, necessity, proportionality, accuracy, non-discrimination, disclosure and independent oversight.

2. No artificial intelligence output shall be treated as conclusive evidence of guilt, dangerousness, credibility, identity, emotion or intention. Affected persons shall have access to the basis of a material decision and a competent human decision-maker.

3. Predictive systems shall not substitute group correlations or proxies for individualized evidence where liberty, protection status or another fundamental right is at stake.

Article 16 Children education health employment and essential services

1. In every action concerning a child, the best interests of the child shall be a primary consideration. Children shall receive age-appropriate information and protection from profiling, addictive design, manipulation and exploitative data practices.

2. Artificial intelligence in education shall support, and not displace, the professional judgement of educators; protect academic freedom and learner privacy; preserve human relationships;

and be accessible, culturally and linguistically appropriate and pedagogically sound.

3. Artificial intelligence in health shall be clinically validated for its intended context, supervised by qualified professionals and governed by patient safety, informed consent where applicable, confidentiality and equitable access.

4. Artificial intelligence in recruitment, work allocation, performance management and dismissal shall be transparent, contestable and subject to worker consultation, collective rights and meaningful human review.

5. Access to housing, credit, insurance, utilities, social protection and other essential services shall not be determined by opaque or discriminatory profiling.

Article 17 Democracy information and culture

1. States Parties shall protect freedom of expression, access to information, cultural diversity, scientific inquiry, artistic freedom and the integrity of democratic processes in the governance of artificial intelligence.

2. Providers and deployers shall take proportionate measures to enable identification of materially deceptive synthetic content, particularly content likely to affect elections, public safety or the rights of others, while safeguarding lawful expression, satire, journalism and anonymity.

3. States Parties shall promote media and artificial intelligence literacy, independent journalism, provenance standards, pluralism and access to reliable public-interest information.

4. Measures addressing harmful content shall be prescribed by law, necessary and proportionate and shall not become a means of censorship or political control.

Article 18 Research innovation and regulatory learning

1. States Parties shall foster responsible research and socially beneficial innovation, including public-interest, open, multilingual and accessibility-oriented artificial intelligence.

2. Regulatory sandboxes and controlled testing environments may be established where they include clear eligibility rules, informed safeguards, independent supervision, time limits and protection from uncompensated harm.

3. Scientific freedom shall be respected. Research exemptions shall not excuse conduct that presents an unreasonable risk of serious harm or violates international law.

---

Chapter IV Duties throughout the artificial intelligence lifecycle

Article 19 Lifecycle governance

1. Every entity that develops, provides or deploys a high-impact artificial intelligence system shall maintain documented governance proportionate to its role and risk, including responsibility at senior level, competent personnel, quality management, security and channels for reporting concerns.

2. The entity shall identify intended and prohibited uses, relevant affected groups, foreseeable failures, dependencies and limitations; establish measurable acceptance criteria; and preserve evidence of compliance.

3. Where adequate mitigation is not reasonably possible, the entity shall not develop, release or continue the activity concerned.

Article 20 Data models and general purpose systems

1. Providers shall document the provenance, composition, lawful basis, quality and known limitations of training, testing and operational data to the extent necessary for accountability and downstream compliance.

2. Providers of general-purpose artificial intelligence models with capabilities capable of posing systemic risk shall evaluate such capabilities before release, adopt proportionate safeguards, report serious incidents and provide downstream actors with sufficient information to use the model responsibly.

3. States Parties shall require heightened measures where a model may materially facilitate cyber operations, biological or chemical harm, large-scale manipulation, loss of control or other severe transboundary harm.

4. Obligations shall take account of open research and open-source development without creating immunity for actors that knowingly enable serious harm.

Article 21 Superintelligence advanced artificial intelligence and existential risk

1. States Parties shall prohibit the development, training, release, deployment or operation of an advanced artificial intelligence system where there are reasonable grounds to consider that it could cause human extinction, the irreversible loss of effective human control, or catastrophic harm of comparable gravity, unless and until the responsible entity demonstrates, through independent evaluation and to the satisfaction of the competent authority, that such risk has been reduced to a demonstrably acceptable level.

2. No person or entity shall create or operate an artificial intelligence system capable of materially improving its own core capabilities, autonomously replicating, acquiring resources, evading oversight or resisting shutdown unless specifically licensed under a stringent national regime consistent with the present Charter.

3. A licence under paragraph 2 shall require, at a minimum: staged authorisation before major training or capability increases; secure containment; controlled access to computing resources, networks, tools and critical infrastructure; independent pre-deployment and continuing evaluations; verifiable alignment and control measures; cybersecurity against theft or misuse; complete technical records; incident reporting; and tested mechanisms for interruption, rollback and safe shutdown.

4. Developers and operators shall maintain meaningful human control and shall not grant an advanced artificial intelligence system authority to reproduce itself, modify its governing objectives, conceal material information from supervisors, initiate irreversible physical actions or control weapons, nuclear systems or other critical infrastructure without specific and revocable human authorisation.

5. Where evidence indicates a credible risk of loss of control, catastrophic harm or circumvention of safeguards, the responsible entity shall immediately suspend the relevant activity, preserve evidence, notify the competent authority and cooperate in containment and investigation. Competent authorities shall have power to order cessation, seizure, isolation or destruction where strictly necessary and proportionate to protect humanity.

6. States Parties shall establish licensing and monitoring of exceptionally large training runs and other capability thresholds designated by the Conference of States Parties; require advance notification and auditable reporting of relevant computing resources; and cooperate in common evaluation standards, emergency protocols and verification arrangements.

7. No competitive, commercial, scientific or national-security interest shall justify proceeding where the requirements of this Article have not been met. The burden of demonstrating compliance shall rest upon the entity proposing the activity.

8. The Conference of States Parties shall review the definitions, capability thresholds and safeguards under this Article at least annually, informed by the Independent Scientific and Human Rights Panel, and may adopt emergency interim measures in response to a newly identified existential risk.

Article 22 Human oversight and professional competence

1. Persons assigned to supervise high-impact artificial intelligence systems shall possess the authority, competence, time, information and organisational support necessary to exercise meaningful judgement.

2. Human oversight shall not be merely formal. Interfaces, workloads, incentives and procedures shall be designed to prevent automation bias and enable effective intervention.

3. States Parties shall promote continuing professional education in artificial intelligence ethics, rights, safety, accessibility, data governance and sector-specific responsibilities.

Article 23 Monitoring incidents and corrective action

1. Providers and deployers shall monitor high-impact systems in operation and shall notify competent authorities without undue delay of a serious incident, near miss or previously unidentified systemic risk.

2. Competent authorities shall maintain interoperable confidential reporting channels and, where public protection so requires, publish anonymized lessons and safety notices.

3. Responsible entities shall investigate root causes, preserve evidence, inform affected persons where appropriate, provide or fund remediation and prevent recurrence.

Article 24 Audit conformity and assurance

1. States Parties shall require independent assurance for high-impact systems at intervals proportionate to risk, including evaluation of performance across relevant populations and real-world conditions.

2. Auditors and assessors shall be competent, impartial and protected from conflicts of interest. They shall have access to the information and systems necessary to reach a reliable conclusion.

3. Certification or conformity assessment shall not replace the continuing responsibility of a provider, deployer or public authority, nor preclude investigation or liability.

---

Chapter V Rights of persons and communities

Article 25 Notice explanation and human review

1. A person materially affected by an artificial intelligence-assisted decision shall have the right to timely and accessible notice, to meaningful information concerning the role of the system and principal factors in the outcome, and to human review by a person authorised to change the decision.

2. Explanations shall be sufficiently specific to permit the person to understand, contest and seek remedy. Legitimate confidentiality shall be protected by measures that do not extinguish these rights.

3. No person shall suffer retaliation for declining an optional artificial intelligence interaction or for requesting human review.

Article 26 Complaint remedy and liability

1. States Parties shall ensure accessible, affordable, timely and effective mechanisms by which persons and communities may lodge complaints and obtain investigation, cessation, correction, compensation, rehabilitation, satisfaction and guarantees of non-repetition as appropriate.

2. Rules on evidence and causation shall not impose an impossible or disproportionate burden on an affected person where relevant information is controlled by another party.

3. States Parties shall provide for civil, administrative and, where appropriate, criminal responsibility proportionate to fault, control, harm and benefit, together with effective penalties and injunctive powers.

4. Collective, representative and public-interest actions shall be available where diffuse or systemic harms would otherwise escape effective remedy.

Article 27 Participation and protection of defenders

1. States Parties shall ensure meaningful, accessible and appropriately resourced participation by affected persons and communities in the formulation, implementation and review of artificial intelligence policy.

2. Particular measures shall support participation by children, persons with disabilities, women and girls, Indigenous Peoples, linguistic minorities, workers, migrants and communities in developing countries.

3. Whistle-blowers, researchers, journalists, human rights defenders and workers who disclose artificial intelligence risks in good faith shall be protected against retaliation, subject to lawful safeguards for the rights of others and public safety.

---

Chapter VI International cooperation peace and equitable development

Article 28 International cooperation and mutual assistance

1. States Parties shall cooperate in good faith to prevent, investigate and remedy transboundary harms; exchange safety information; coordinate emergency responses; and promote compatible standards and regulatory approaches.

2. Cooperation shall respect human rights, data protection, due process and the sovereign equality of States. Assistance shall not be conditioned on the adoption of weaker protections or on inequitable access to data, infrastructure or markets.

3. States Parties shall provide the widest measure of lawful mutual assistance in enforcement, including preservation of evidence and service of process, and shall establish secure channels among competent authorities.

Article 29 Capacity building and global equity

1. States Parties and international organisations in a position to do so shall provide predictable, needs-based assistance for legal and regulatory capacity, public-interest compute, data stewardship, cybersecurity, research, education, local-language resources and independent oversight.

2. Capacity-building priorities shall be determined with recipient countries and communities, support local institutions and expertise, include women and underrepresented groups, and avoid technological dependency or extractive data practices.

3. States Parties shall promote fair access to the benefits of artificial intelligence and shall consider appropriate mechanisms for financing global public goods, including contributions proportionate to the scale and risk of advanced artificial intelligence activity.

Article 30 International peace security and humanitarian law

1. Artificial intelligence shall not be developed or used in a manner inconsistent with the obligation to settle international disputes by peaceful means, the prohibition on the threat or use of force, international humanitarian law or international criminal law.

2. States Parties shall maintain human command and accountability over military and security applications and shall conduct legal review of new weapons, means or methods of warfare involving artificial intelligence.

3. States Parties shall cooperate to prevent artificial intelligence from facilitating terrorism, trafficking, cyberattacks, proliferation of weapons of mass destruction or other threats to peace, while ensuring that measures taken comply with international law.

4. The present Charter shall not be interpreted as authorizing any activity otherwise prohibited by international law.

---

Chapter VII International institutions and oversight

Article 31 Conference of States Parties

1. A Conference of States Parties is hereby established to review implementation, adopt protocols and technical guidance, facilitate cooperation and assistance, and consider emerging risks and beneficial uses.

2. Each State Party shall have one vote. The Conference shall seek consensus and, where consensus cannot be reached, may take decisions in accordance with rules of procedure adopted by a two-thirds majority.

3. The Conference shall ensure structured participation, without decision-making authority, by international organisations, civil society, academia, technical communities, workers, industry and representatives of affected groups.

Article 32 International Artificial Intelligence Council

1. An International Artificial Intelligence Council is hereby established as the standing implementation body of the present Charter, accountable to the Conference of States Parties and cooperating with the competent organs and agencies of the United Nations.

2. The Council shall facilitate regulatory cooperation, maintain a global register of designated high-impact and systemic-risk systems, coordinate alerts and mutual assistance, support capacity-building and prepare implementation guidance.

3. Membership shall reflect equitable geographical distribution, gender balance, legal and technical expertise, and representation of developing countries. Members shall act independently and disclose conflicts of interest.

Article 33 Independent Scientific and Human Rights Panel

1. The Conference shall be advised by an independent multidisciplinary Panel on Artificial Intelligence, Science and Human Rights, building upon relevant United Nations mechanisms and avoiding unnecessary duplication.

2. The Panel shall publish periodic assessments of capabilities, benefits, risks, incidents, environmental effects and distributional impacts; identify scientific uncertainty; and recommend proportionate responses.

3. The Panel shall use transparent methods, protect legitimate confidential information and include expertise from all regions and from the social sciences, humanities, law, education, disability, gender, labour, environment and affected communities as well as technical fields.

Article 34 National reports and international review

1. Within two years after the present Charter enters into force for a State Party, and every four years thereafter, that State shall report on measures taken, outcomes achieved, significant incidents, enforcement, remedies, public participation and assistance provided or received.

2. Reports shall be public and accessible, subject to narrowly tailored redactions. The Council shall conduct a constructive, non-selective and evidence-based review and may issue recommendations.

3. The review shall take account of national capacities and shall link identified needs to technical and financial assistance.

Article 35 Communications inquiries and urgent measures

1. A State Party may recognise the competence of an independent compliance committee established by the Conference to receive communications from affected persons or groups after reasonably available domestic remedies have been exhausted, unless such remedies are unduly prolonged or ineffective.

2. Where reliable information indicates grave or systematic violations, the committee may, with due process and the cooperation of the State concerned, undertake an inquiry and recommend protective or remedial measures.

3. In a situation presenting an imminent risk of widespread and irreversible harm, the Council may issue a public alert and recommend temporary precautionary measures, without prejudice to the powers of other competent international bodies.

---

Chapter VIII Final provisions

Article 36 Relationship with other instruments

1. The present Charter shall be interpreted consistently with the Charter of the United Nations and applicable international law.

2. Nothing in the present Charter shall affect rights and obligations arising from instruments that provide equal or greater protection. States Parties shall promote coherence among international, regional and national artificial intelligence regimes.

3. Specialized protocols may be adopted concerning matters including autonomous weapons, frontier model safety, health, education, labour, climate, data governance and criminal cooperation.

Article 37 Settlement of disputes

1. States Parties shall seek to settle disputes concerning the interpretation or application of the present Charter by negotiation, inquiry, mediation, conciliation, arbitration, judicial settlement, resort to regional arrangements or other peaceful means of their own choice.

2. A State Party may, upon signature, ratification or accession, declare that it recognises as compulsory, in relation to any State accepting the same obligation, arbitration under procedures adopted by the Conference or submission to the International Court of Justice.

3. Nothing in this Article shall impair the functions and powers of the organs of the United Nations under the Charter of the United Nations.

Article 38 Signature ratification acceptance approval and accession

1. The present Charter shall be open for signature by all States at United Nations Headquarters in New York.

2. The present Charter is subject to ratification, acceptance or approval. It shall remain open for accession by any State. Instruments shall be deposited with the Secretary-General of the United Nations.

3. Regional integration organisations may become Parties within the limits of their competence under conditions determined by the negotiating conference.

Article 39 Entry into force and provisional application

1. The present Charter shall enter into force on the ninetieth day after the deposit of the fortieth instrument of ratification, acceptance, approval or accession.

2. For each State becoming a Party thereafter, it shall enter into force on the ninetieth day after deposit of that State’s instrument.

3. A State may declare that it will apply the present Charter provisionally, in whole or in part, pending its entry into force for that State.

Article 40 Amendments and review

1. Any State Party may propose an amendment. The Conference shall seek consensus and may adopt an amendment by a two-thirds majority of States Parties present and voting where consensus cannot be reached.

2. An amendment shall enter into force for accepting States under procedures adopted by the negotiating conference, with enhanced safeguards for amendments affecting institutional powers or fundamental rights.

3. The Conference shall conduct a comprehensive review five years after entry into force and at intervals not exceeding five years thereafter, taking account of scientific evidence, lived experience and developments in international law.

Article 41 Reservations withdrawal and depositary

1. No reservation incompatible with the object and purpose of the present Charter shall be permitted.

2. A State Party may withdraw by written notification to the Depositary. Withdrawal shall take effect one year after receipt and shall not release the State from obligations arising from acts or omissions occurring before that date or from obligations under other international law.

3. The Secretary-General of the United Nations shall be the Depositary of the present Charter.

Article 42 Authentic texts

1. The Arabic, Chinese, English, French, Russian and Spanish texts of the present Charter shall be equally authentic.

2. The Depositary shall promote accessible formats and authoritative translations into additional languages, including sign languages and formats accessible to persons with disabilities.

Source framework and alignment

This model text draws upon the following instruments as a source framework. The references indicate conceptual alignment; they do not imply endorsement of this draft by the issuing organisations.

Instrument Relevant alignment Official reference Charter of the United Nations Purposes and principles, sovereign equality, peaceful settlement, international cooperation and institutional form.

https://www.un.org/en/ about-us/un-charter/full-text Universal Declaration of Human Rights and core human rights treaties Human dignity, equality, privacy, expression, participation, due process, work, education and effective remedy.

https://www.un.org/en/ about-us/universal-declaration-of-human- Instrument Relevant alignment Official reference rights UNESCO Recommendation on the Ethics of Artificial Intelligence 2021 Proportionality and do no harm, safety, fairness, privacy, human oversight, transparency, accountability, sustainability and literacy.

https:// unesdoc.unesco.org/ark:/ 48223/pf0000381137 United Nations General Assembly resolution 78 265 International consensus on safe, secure and trustworthy AI systems, human rights, sustainable development and closing digital divides.

https:// digitallibrary.un.org/ record/4043244 Global Digital Compact 2024 Inclusive digital cooperation, data governance, human rights, capacity-building and international governance of AI for the benefit of humanity.

https://www.un.org/ global-digital-compact/ sites/default/files/2024- 09/ Global%20Digital%20Com pact%20- %20English_0.pdf Convention on the Rights of Persons with Disabilities Accessibility, universal design, reasonable accommodation, participation and equality.

https://www.ohchr.org/ en/instruments-mechanisms/ instruments/convention-rights-persons-disabilities Convention on the Rights of the Child Best interests of the child, development, participation, privacy, protection and education.

https://www.ohchr.org/ en/instruments-mechanisms/ instruments/convention-rights-child Council of Europe Framework Convention on Artificial Intelligence Lifecycle duties, risk and impact assessment, transparency, accountability, remedies, democracy and the rule of law.

https://www.coe.int/en/ web/artificial-intelligence/the-framework-convention-on-artificial-intelligence

Drafting note: Terms and institutional arrangements in this model text would require legal and

diplomatic refinement in an intergovernmental process, including decisions on the relationship with existing United Nations bodies, financing, privileges and immunities, verification powers and the legal form of adoption.