Convention On Artificial Intelligence, Human Rights, Democracy And The Rule of Law

The Council of Europe has on the 17^th of May 2024, finalised a first of its kind, legally binding draft international treaty on Artificial Intelligence and Human Rights, Democracy and the Rule of Law (the “Convention”). The drafters of the Convention, conscious of the developments brought about by Artificial Intelligence (“AI”), address the risks of discrimination in digital contexts, while considering the manner in which AI systems could possibly undermine human dignity and individual autonomy, human rights, democracy and the rule of law, if misused. For instance, the COMPAS algorithm used in the U.S. criminal justice system¹Julia Angwin, Jeff Larson, Surya Mattu and Lauren Kirchner, “Machine Bias”, ProPublica May 23, 2016, < https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing> has been found to be biased against African American defendants, potentially leading to unjust sentencing and exacerbating racial inequalities. Similarly, Amazon’s AI recruiting tool was scrapped after it was discovered to be biased against women²Jeffrey Dastin, “Insight – Amazon scraps secret AI recruiting tool that showed bias against women”, Reuters, Oct 2018, < https://www.reuters.com/article/us-amazon-com-jobs-automation-insight-idUSKCN1MK08G/> , highlighting how AI can perpetuate existing biases if not carefully managed.

The legal framework established pursuant to the Convention, sets out common general principles and rules governing the activities within the lifecycle of AI systems, that effectively preserves shared values and harnesses the benefits of AI for the promotion of these values in a manner conducive to responsible innovation.

The provisions of the Convention imposes, upon the 46 member states of the Council of Europe (the “Member States”), stringent standards for developing, deploying, and using AI technologies. It also sets outs general obligations to-

1. adopt or maintain measures to ensure that the activities of AI systems are consistent with obligations to protect human rights, as enshrined in applicable international law and in the Member States’ domestic legislation;
2. seek to ensure that AI systems are not used to undermine the integrity, independence and effectiveness of democratic institutions and processes, including the principle of the separation of powers, respect for judicial independence and access to justice; and

• implement legal and regulatory measures that ensure AI transparency, accountability, fairness, equality non-discrimination, and respect to privacy rights and data protection.

Additionally, the Member States must have in place effective procedural guarantees, safeguards and rights, in accordance with the applicable international and domestic law to ensure the availability of accessible and effective remedies for violations of human rights resulting from the activities of AI systems.

In order to ensure its effective implementation, the Convention also establishes a follow-up mechanism in the form of a Conference of the Parties and mandates Member States to appoint an independent oversight mechanism to oversee compliance with the Convention, raise awareness, stimulate an informed public debate, and carry out multistakeholder consultations on how AI technology should be used.

The Convention, although in its draft form, has not fallen short of criticism. The Commissioner for Human Rights has raised its concern of the lack of positive obligations to create a legal and regulatory framework that effectively protects individuals from all human rights violations, whether these are committed by public or private sector actors as well as its lack of explicit safeguard of the right to an effective remedy, including the right to human review of automated decisions, aligned with Article 13 of the European Convention on Human Rights. The European Data Protection Supervisor (“EDPS”) in a statement issued in March 2024, has raise various shortcomings including the fact that the proposed position are too generic and are largely declarative nature. This would inevitably lead to divergent application of the Convention, thereby undermining legal certainty, and more generally its added value. However, the matter which causes the EDPS most unease is the absence of what it calls “red lines” which from the outset prohibit AI applications posing unacceptable levels of risk.

Finally, the Convention, which his set to be signed in September of this year, will certainly complement the EU AI Act. From a reading of the two frameworks, it is evident that the Convention and the EU AI Act share a risk-based approach and similar core principles. Despite the Convention’s general wording and broad exceptions, the EU AI Act, adopted on May 21, 2024, is arguably the most comprehensive and prescriptive set of AI standards within the European Union. The Convention will, nonetheless, form the foundation of AI regulation in the Council of Europe and it is expected that the European Court of Human Rights (ECtHR) will reference the Convention when interpreting alleged breaches of the European Convention on Human Rights (ECHR). This means that the principles and provisions outlined in the Convention will inevitably influence the European Court of Human Right’s judgments, potentially leading to new legal standards and protections regarding the misuse of AI.

The Draft Convention can be accessed through this link.