Convention on AI (draft January 2023)
February 2023
Conventions and Treaties
Revised Zero Draft [Framework] Convention on
Artificial Intelligence, Human Rights, Democracy, and the Rule of Law
COMMITTEE ON ARTIFICIAL INTELLIGENCE (CAI)
January 2023
This document was prepared by the Chair of the CAI with the support of the Secretariat to serve as the basis for the drafting of the [Framework] Convention on Artificial Intelligence, Human Rights, Democracy and the Rule of Law. It does not reflect the final outcome of negotiations in the Committee.
Contents
TogglePreamble
The member States of the Council of Europe and the other signatories hereto,
Considering that the aim of the Council of Europe is to achieve greater unity between its members, based in particular on respect for human rights and fundamental freedoms, democracy and the rule of law;
Recognising the value of fostering cooperation between the Parties to this Convention;
Conscious of the accelerating developments in science and technology and the profound changes brought about by the application of artificial intelligence systems through the provision of new tools, which have the potential to promote human prosperity as well as individual and social well-being by enhancing progress and innovation;
Concerned by the risk that certain uses of some artificial intelligence systems also have the potential for unduly interfering with the exercise of human rights and fundamental freedoms, undermining democracy and violating the rule of law;
Conscious of the fact that human rights, democracy and the rule of law are inherently interwoven and convinced of the need to establish, as a matter of priority, a common legal framework establishing fundamental principles and rules governing design, development, and application of artificial intelligence systems which would effectively preserve the shared common values and at the same time be conducive to innovation;
Conscious also of the fact that in view of its framework character the Convention may be supplemented by further binding or non-binding instruments to address challenges relating to the application of artificial intelligence systems in specific sectors;
Welcoming efforts undertaken by other international and supranational Organisations and fora, which further advance international understanding and cooperation in this area;
Mindful of the need to ensure respect for human rights as enshrined in the 1950 Council of Europe Convention for the Protection of Human Rights and Fundamental Freedoms and its protocols, the 1966 United Nations International Covenant on Civil and Political Rights and other applicable international human rights treaties while taking into due account various economic, security and other interests in the development and use of artificial intelligence systems;
Underlining that the present Convention is intended to complement those conventions in order to fill in any legal gaps in view of the specific challenges raised by design, development and application of artificial intelligence systems;
Mindful also of the right to privacy and the protection of personal data, as conferred, for example, by the 1981 Council of Europe Convention for the Protection of Individuals with regard to Automatic Processing of Personal Data and amending protocols;
Affirming the commitment of Parties to protecting human rights, democracy and the rule of law and opposing unlawful and irresponsible use of artificial intelligence technologies, which is not in line with their shared values,
Have agreed as follows:
Chapter I: General provisions
Article 1 – Purpose and object
1. This Convention establishes certain fundamental principles, rules and rights aimed at ensuring that design, development and application of artificial intelligence systems is fully consistent with respect for human rights, the functioning of democracy and the observance of rule of law.
2. Each Party shall take the necessary measures in its domestic legislation to give effect to the principles, rules and rights set out in this Convention.
3. In order to ensure effective implementation of its provisions by its Parties, this Convention establishes a follow-up mechanism.
Article 2 – Definitions
For the purposes of this Convention:
a. “artificial intelligence system” means any algorithmic system or a combination of such systems that, as defined herein and in the domestic law of each Party, uses computational methods derived from statistics or other mathematical techniques to carry out functions that are commonly associated with, or would otherwise require, human intelligence and that either assists or replaces the judgment of human decision-makers in carrying out those functions. Such functions include, but are not limited to, prediction, planning, classification, pattern recognition, organisation, perception, speech/sound/image recognition, text/sound/image generation, language translation, communication, learning, representation, and problem- solving;
b. “lifecycle” means all phases of existence of an artificial intelligence system between its design and decommissioning;
c. “artificial intelligence provider” means any natural or legal person, public authority or other body that develops an artificial intelligence system with a view to putting it into service/commissioning it;
d. “artificial intelligence user” means any natural or legal person, public authority or other body using an artificial intelligence system in their own name or under their authority;
e. “artificial intelligence subject” means any natural or legal person whose human rights and fundamental freedoms or connected legal rights guaranteed under applicable domestic law or international law are impacted through the application of an artificial intelligence system, including by decisions made or substantially informed by the application of such system.
Article 3 – Principle of non-discrimination
The implementation of the provisions of this Convention by the Parties shall be secured without discrimination on any ground such as sex, gender, sexual orientation, race, colour, language, age, religion, political or any other opinion, national or social origin, association with a national minority, property, birth, state of health, disability or other status, or based on a combination of one or more of these grounds.
Article 4 – Scope
1. This Convention shall apply to design, development and application of artificial intelligence systems that are used in a context involving issues relating to the respect for human rights, the functioning of democracy and the observance of rule of law, as described in Articles 5 (a), 7 and 8 (a).
2. This Convention shall apply to such systems throughout their lifecycle, regardless of whether these activities are undertaken by public or private actors.
3. This Convention shall not apply to design, development and application of artificial intelligence systems used for purposes related to national defense.
Chapter II: Application of artificial intelligence systems by public authorities
Article 5 – Obligations relating to public authorities
Each Party shall, within its respective jurisdiction, ensure that:
a. the application of an artificial intelligence system substantially informing decision-making by a public authority in the exercise of its function, or any private entity acting on its behalf, is fully compatible with its obligations to respect human rights and fundamental freedoms as guaranteed under its domestic law or under any relevant applicable international law;
b. any interference with human rights and fundamental freedoms by a public authority or any private entity acting on its behalf resulting from such application of an artificial intelligence system is compatible with core values of democratic societies, in accordance with the law and necessary in a democratic society in pursuit of a legitimate public interest.
Article 6 – Requirements regarding respect for human rights
Each Party shall take such measures as are aimed at minimising and, to the extent possible, preventing any unlawful harm or damage to human rights and fundamental freedoms, which could result from the inappropriate application of artificial intelligence systems by public authorities.
Article 7 – Requirements regarding respect for democratic institutions and the rule of law
Each Party shall take all necessary measures to preserve the integrity of democratic institutions and processes and ensure the respect for the rule of law and proper administration of justice in the context of application of an artificial intelligence system. To this end, a thorough assessment of the necessity, proportionality and potential risks shall be carried out by a competent domestic authority both prior and during the application of an artificial intelligence system to the elaboration or revision of laws and policies governing the respective roles of the executive authority, other democratic institutions and the judiciary.
Chapter III: Application of artificial intelligence systems in provision of goods, facilities and services
Article 8 – Obligations relating to public and private actors
Each Party shall, within its respective jurisdiction, ensure that:
a. the application of an artificial intelligence system in provision of goods, facilities and services in essential areas, such as but not restricted to, health, family care, housing, energy consumption, transport, food supply, education, employment, finance, environmental protection, digital information, media and communication is fully compatible with its domestic law and any applicable international law insofar as these require relevant public and private actors to respect human rights and fundamental freedoms;
b. effective guidance is provided to relevant public and private actors on how to prevent and mitigate any adverse impacts of the application of an artificial intelligence system on the enjoyment of human rights and fundamental freedoms, the functioning of democracy and the observance of the rule of law in their operations.
Article 9 – Preservation of individual freedom, human dignity and autonomy
Each Party shall take the necessary measures to preserve individual freedom, human dignity and autonomy and in particular the ability to reach informed decisions free from undue influence, manipulation or detrimental effects which may adversely affect the right to freedom of expression and assembly, democratic participation and the exercise of other relevant human rights and fundamental freedoms as a result of the inappropriate application of an artificial intelligence system.
Article 10 – Access to public debate and inclusive democratic processes
Each Party shall take the necessary measures to ensure that all interested parties, groups and individuals enjoy equal and fair access to public debate and inclusive democratic processes. In doing so, it shall take due account of the relevant implications of technological developments and the role of public and private entities using artificial intelligence systems that may help shape public debate.
Article 11 – Preservation of public health and the environment
Each party shall take the necessary measures to preserve public health and the environment in the context of the application of an artificial intelligence system.
Chapter IV: Fundamental principles of design, development and application of artificial intelligence systems
Article 12 – Principle of Equality and Anti-discrimination
Each Party shall, within its jurisdiction and in accordance with its domestic law, ensure that the design, development and application of artificial intelligence systems respect the principle of equality, including gender equality and rights related to discriminated groups and individuals in vulnerable situations.
Article 13 – Principle of Privacy and Personal Data Protection
Each Party shall, within its jurisdiction, ensure that as regards the design, development and application of artificial intelligence systems:
a. the privacy of individuals is protected including through applicable domestic and international personal data protection and data governance laws and standards;
b. relevant data protection law and principles are applied;c. appropriate guarantees and safeguards have been put in place for data subjects.
Article 14 – Principle of Accountability, Responsibility and Legal Liability
Each Party shall, within its jurisdiction, take measures necessary to ensure accountability, responsibility and legal liability for any unlawful harm or damage to human rights and fundamental freedoms resulting from the application of artificial intelligence systems.
Article 15 – Principle of Transparency and Oversight
Each Party shall, within its jurisdiction and in accordance with its domestic law, ensure that adequate oversight mechanisms as well as transparency and auditability requirements tailored to the specific risks arising from the context in which the artificial intelligence systems are applied are in place.
Article 16 – Principle of Safety
Each Party shall, within its jurisdiction and in accordance with its domestic law, ensure that adequate safety, security, data quality, data integrity, data security, cybersecurity and robustness requirements are in place for design, development and application of artificial intelligence systems.
Article 17 – Principle of Safe Innovation
With a view to encouraging research and fostering innovation each Party shall, within its jurisdiction and in accordance with its domestic law, provide for the establishment of controlled regulatory environment for testing artificial intelligence systems for artificial intelligence providers and users under the supervision of its competent authorities.
Article 18 – Public consultation and additional measures
1. Parties shall strive to ensure that fundamental questions raised by the design, development and application of artificial intelligence systems are the subject of appropriate public discussion and multi-stakeholder consultation in the light, in particular, of relevant social, economic, ethical and legal implications.
2. Parties shall encourage and promote digital literacy and digital skills for all segments of the population.
Chapter V: Measures and Safeguards ensuring Accountability and Redress
Article 19 – Measures ensuring the availability of redress
Each Party shall, within its jurisdiction and in accordance with its domestic law, take measures ensuring the availability of redress for any unlawful harm or damage to human rights and fundamental freedoms resulting from the application of artificial intelligence systems through:
a. setting up a system, in accordance with its domestic law, whereby the relevant usage of the system is recorded and, where appropriate, communicated to the artificial intelligence subjects concerned;
b. guaranteeing that such communication contains sufficient information for an effective possibility of contesting the application of the system or challenging the decision(s) affecting the artificial intelligence subject’s rights and freedoms insofar as the use of such system is concerned;
c. making available effective redress mechanisms.
Article 20 – Additional Procedural Safeguards
1. Each Party shall, within its jurisdiction and in accordance with its domestic law, ensure that where an artificial intelligence system substantially informs or takes decision(s) affecting human rights and fundamental freedoms there is a right to human review of the decisions.
2. Each Party shall, within its jurisdiction and in accordance with its domestic law, ensure that any person has the right to know that one is interacting with an artificial intelligence system rather than with a human and, where appropriate, shall provide for the option of interacting with a human in addition to or instead of such system.
3. Each Party shall take appropriate measures to ensure effective access to the safeguards and rights mentioned in Articles 19 and 20 to persons with disabilities.
Article 21 – Restrictions
Restrictions on the exercise of the rights specified in Articles 19 and 20 may only be provided for by law where necessary and proportionate in a democratic society in the pursuit of a legitimate public interest.
Article 22 – Relationship with other legal instruments
Nothing in the present Convention shall be construed as limiting or derogating from any of the human rights and fundamental freedoms as well as legal rights which may be guaranteed under the laws of any Party or under any other agreement to which it is a Party.
Article 23 – Wider protection
None of the provisions of this Convention shall be interpreted as limiting or otherwise affecting the possibility for a Party to grant a wider measure of protection than is stipulated in this Convention.
Chapter VI: Assessment and Mitigation of Risks and Adverse Impacts
Article 24 – Risk and impact management framework
1. Each Party shall, in accordance with its domestic law, provide effective guidance to relevant artificial intelligence providers and users on how to identify, assess, prevent and mitigate risks and adverse impacts resulting from the application of an artificial intelligence system in relation to the enjoyment of human rights, the functioning of democracy and the observance of rule of law.
2. Such guidance shall ensure that the risk and impact assessments and the application of the related prevention and mitigation measures:
a. contain proportionate requirements targeted at the specific contexts of application of an artificial intelligence system which pose increased risks for human rights, democracy and the rule of law;
b. integrate the perspective of artificial intelligence subjects whose rights may be potentially impacted through the operation of the system;
c. require the recording and due consideration of adverse impacts resulting from application of enhanced artificial intelligence systems in relation to the enjoyment of human rights, the functioning of democracy and the observance of rule of law;
d. are carried out iteratively, at appropriate rate, throughout the lifecycle of such system.
3. Each Party shall, in accordance with its domestic law, provide for the possibility of imposing a moratorium or ban on certain applications of artificial intelligence systems where such measure is considered appropriate and necessary by the competent domestic authority.
Article 25 – Obligations of artificial intelligence providers and users
Each Party shall ensure that artificial intelligence providers and users, where applicable:
a. apply sufficient preventive and mitigating measures resulting from the application of the aforementioned risk and impact management framework;
b. document and keep the records of the respective processes.
Article 26 – Training
Each Party shall ensure that competent supervisory authorities, artificial intelligence providers and users receive appropriate training in connection with the operation of the risk and impact management framework.
Chapter VII: Follow-up mechanism and cooperation
Article 27 – Consultation of the Parties
1. Parties shall consult periodically with a view to:
a. making proposals to facilitate or improve the effective use and implementation of this Convention, including the identification of any problems and the effects of any declaration made under this Convention;
b. making proposals for the amendment of this Convention in accordance with Article 31;
c. formulating their opinion on any proposal for the amendment of this Convention which is referred to them in accordance with Article 31, paragraph 3;
d. expressing an opinion on any question concerning the interpretation and application of this Convention and facilitating the exchange of information on significant legal, policy or technological developments.
2. The Consultation of the Parties shall be convened by the Secretary General of the Council of Europe whenever (s)he finds it necessary and in any case when a majority of the Parties or the Committee of Ministers request its convocation.
3. The Consultation of the Parties shall adopt its own rules of procedure.
4. Parties shall be assisted by the Secretariat of the Council of Europe in carrying out their functions pursuant to this article.
5. A contracting Party which is not a member of the Council of Europe shall contribute to the financing of the Consultation of the Parties in a manner to be decided by the Committee of Ministers upon consultation of that Party.
Article 28 – International co-operation
1. Parties shall, as appropriate, co-operate to the fullest extent in the realisation of the purpose of this Convention.
2. Parties shall, on a regular basis, exchange information between them concerning design, development and application of artificial intelligence systems which they assess as posing a risk to the enjoyment of human rights, the functioning of democracy and the observance of rule of law.
Article 29 – National supervisory authorities
1. Parties shall establish or designate national supervisory authorities tasked, in particular, with overseeing and supervising compliance with the requirements of the risk and impact assessment of artificial intelligence systems, in accordance with Articles 24 and 25.
2. Parties shall ensure the national supervisory authorities have sufficient resources and properly trained personnel to carry out their activities.
3. The national supervisory authorities shall be independent and impartial in the exercise of their duties.
Chapter VIII: Final provisions
Article 30 – Effects of the Convention
1. If two or more Parties have already concluded an agreement or treaty on the matters dealt with in this Convention or have otherwise established their relations on such matters, or should they in future do so, they shall also be entitled to apply that agreement or treaty or to regulate those relations accordingly. However, where Parties establish their relations in respect of the matters dealt with in the present Convention other than as regulated therein, they shall do so in a manner that is not inconsistent with the Convention’s objective and principles.
2. Parties which are members of the European Union shall, in their mutual relations, apply Community and European Union rules in so far as there are Community or European Union rules governing the particular subject concerned and applicable to the specific case, without prejudice to the object and purpose of the present Convention and without prejudice to its full application with other Parties.
3. Nothing in this Convention shall affect other rights, obligations and responsibilities of a Party and individuals under international law.
Article 31 – Amendments
1. Amendments to this Convention may be proposed by any Party, the Committee of Ministers of the Council of Europe or the Consultation of the Parties.
2. Any proposal for amendment shall be communicated by the Secretary General of the Council of Europe to the Parties.
3. Moreover, any amendment proposed by a Party, or the Committee of Ministers, shall be communicated to the Consultation of the Parties, which shall submit to the Committee of Ministers its opinion on the proposed amendment.
4. The Committee of Ministers shall consider the proposed amendment and any opinion submitted by the Consultation of the Parties and may approve the amendment.
5. The text of any amendment approved by the Committee of Ministers in accordance with paragraph 4 shall be forwarded to the Parties for acceptance.
6. Any amendment approved in accordance with paragraph 4 shall come into force on the thirtieth day after all Parties have informed the Secretary General of their acceptance thereof.
Article 32 – Dispute settlement
In the event of a dispute between Parties as to the interpretation or application of this Convention which cannot be resolved by the Consultation of the Parties, as provided for in Article 27 , paragraph 1, d, they shall seek a settlement of the dispute through negotiation or any other peaceful means of their choice, including submission of the dispute to an arbitral tribunal whose decisions shall be binding upon the Parties to the dispute, or to the International Court of Justice, as agreed upon by the Parties concerned.
Article 33 – Signature and entry into force
1. This Convention shall be open for signature by the member States of the Council of Europe, by the European Union and by non-member States which have participated in its elaboration.
2. This Convention is subject to ratification, acceptance or approval. Instruments of ratification, acceptance or approval shall be deposited with the Secretary General of the Council of Europe.
3. This Convention shall enter into force on the first day of the month following the expiration of a period of three months after the date on which six Signatories, including at least four member States of the Council of Europe, have expressed their consent to be bound by the Convention in accordance with the provisions of paragraph 2.
4. In respect of any Signatory which subsequently expresses its consent to be bound by it, the Convention shall enter into force on the first day of the month following the expiration of a period of three months after the date of the expression of its consent to be bound by the Convention in accordance with the provisions of paragraph 2.
Article 34 – Accession
1. After the entry into force of this Convention, the Committee of Ministers of the Council of Europe, after consulting with and obtaining the unanimous consent of the Parties to the Convention, may invite any State which is not a member of the Council of Europe and which has not participated in its elaboration to accede to this Convention. The decision shall be taken by the majority provided for in Article 20.d of the Statute of the Council of Europe and by the unanimous vote of the representatives of the Parties entitled to sit on the Committee of Ministers.
2. In respect of any State acceding to the Convention under paragraph 1 above, the Convention shall enter into force on the first day of the month following the expiration of a period of three months after the date of deposit of the instrument of accession with the Secretary General of the Council of Europe.
Article 35 – Territorial application
1. Any State or the European Union may, at the time of signature or when depositing its instrument of ratification, acceptance, approval or accession, specify the territory or territories to which this Convention shall apply.
2. Any Party may, at a later date, by a declaration addressed to the Secretary General of the Council of Europe, extend the application of this Convention to any other territory specified in the declaration. In respect of such territory the Convention shall enter into force on the first day of the month following the expiration of a period of three months after the date of receipt of the declaration by the Secretary General.
3. Any declaration made under the two preceding paragraphs may, in respect of any territory specified in such declaration, be withdrawn by a notification addressed to the Secretary General of the Council of Europe. The withdrawal shall become effective on the first day of the month following the expiration of a period of three months after the date of receipt of such notification by the Secretary General.
Article 36 – Reservations
No reservation may be made in respect of any provision of this Convention.
Article 37 – Denunciation
1. Any Party may, at any time, denounce this Convention by means of a notification addressed to the Secretary General of the Council of Europe.
2. Such denunciation shall become effective on the first day of the month following the expiration of a period of three months after the date of receipt of the notification by the Secretary General.
Article 38 – Notification
The Secretary General of the Council of Europe shall notify the member States of the Council of Europe, the non-member States enjoying observer status with the Council of Europe, the European Union, the non-member States which have participated in the elaboration of this Convention, and any State having been invited to sign this Convention in accordance with the provisions of Article 33, or having been invited to accede to it in accordance with the provisions of Article 34, of:
a. any signature;
b. the deposit of any instrument of ratification, acceptance or approval;
c. any date of entry into force of this Convention in accordance with Article 33, paras. 3 and 4, and Article 34, para. 2;
d. any amendment adopted in accordance with Article 31 and the date on which such an amendment enters into force;
e. any denunciation made in pursuance of Article 37 ;
f. any other act, notification or communication relating to this Convention.
In witness whereof the undersigned, being duly authorised thereto, have signed this Convention.
Done in [place], this … day of [month] 202[4], in English and in French, both texts being equally authentic, in a single copy which shall be deposited in the archives of the Council of Europe. The Secretary General of the Council of Europe shall transmit certified copies to each member State of the Council of Europe, to the non-member States which enjoy observer status with the Council of Europe, to the European Union and to any State invited to sign or accede to this Convention.