In March 2024, the Council of Europe decided upon a Convention on AI, intended to be the first binding agreement on AI on an international level.[1] The agreement aims to protect human rights, democracy and the rule of law from harmful AI applications.[2] The international convention's primary function is to bind states and other international organisations to become signing parties of the convention with the obligation to integrate the rules and principles into their legal systems. The Convention on AI does not place obligations on individuals and private organisations. It can be seen as a guiding framework at an international level and can impact states when they ratify the convention and translate the international rules into local laws and regulations.[3]

The draft concerning equality, including gender equality and non-discrimination, refers to already existing human rights law at international and domestic levels, including constitutional law and jurisprudence, which should provide the basis for ensuring rights are guaranteed in the context of AI systems[4] The Explanatory Report also explains the drafters' reflection ‘on the real and well-documented risk of bias that can constitute unlawful discrimination arising from the activities within the lifecycle of artificial intelligence systems’.[5] Parties have ‘to consider appropriate regulatory, governance, technical or other solutions’ for the different ways where bias can ‘be incorporated into artificial intelligence systems at various stages throughout their lifecycle’.[6]  It is underlined that it is not enough to stop ‘requiring that a person is not treated less favourably’ but that ‘structural and historical inequalities’ should also be overcome.[7]

Art. 17 of the Convention addresses how the principles of non-discrimination are implemented as follows: ‘The implementation of the provisions of this Convention by the Parties shall be secured without discrimination on any ground, in accordance with their international human rights obligations.’[8]

On 21 March 2024, the United Nations General Assembly adopted the first resolution on the topic of AI.[9] The Resolution recognises AI's dangers by stating that the design, development and deployment of AI systems without consistency with international law pose a risk to reaching the Sustainable Development Goals, widen the digital gap between countries, reinforce inequalities that lead to bias and discrimination, undermine information integrity, and pose risks for the protection of human rights and fundamental freedoms. Furthermore, the Resolution urges a ‘global consensus on safe, secure and trustworthy artificial intelligence systems’. [10] It calls upon the member states to ‘close the gender digital divide’ and ‘to mainstream a disability, gender and racial equality perspective in policy decisions and the framework that guide them’.[11]

The UN encourages member states and the private sector to develop and support approaches which work towards safe, secure and trustworthy AI systems.[12] It further encourages the private sector ‘to adhere to applicable international and domestic laws and act in line with the United Nations Guiding Principles on Business and Human Rights’.[13] This direct mention of the private sector is understood as a reinforcement of the central role of the private sector in implementing human rights. Companies need to ensure their products align with human rights laws.[14] However, the Resolution does not set international guidelines but promotes the development and implementation of domestic regulatory approaches and frameworks in line with national policies.[15]

The common ground that could be found in the UN Resolution is relatively minor. The UN Resolution commits member states to ensure a safe environment for AI systems to comply with international human rights obligations. The document underlines that states are aware of the dangers AI can pose to individuals. However, from a legal perspective, the Resolution is largely symbolic, as it involves no legal obligations.[16]

[1] Angela Müller and Matthias Spielkamp, ‘Europarat: KI-Konvention wird Menschenrechte nicht angemessen schützen‘ Algorithm Watch (18 March 2024) <algorithmwatch.org/de/ki-konvention-ungenugend/>.

[2] Müller (n 302).

[3] Osman Gazi Güçlütürk, ‘Understanding the Council of Europe’s Draft Framework Convention on AI, Human Rights, Democracy, and Rule of Law’ Holistic AI (17 January 2024) <holisticai.com/blog/europe-committee-artificial-intelligence-draft-framework-convention#:~:text=The%20main%20purpose%20of%20the,(1)%20of%20the%20DFC>.

[4] Council of Europe (n 25) para 71, 74.

[5] Ibid, para 75.

[6] Ibid, para 75.

[7] Ibid, para 77.

[8] Council of Europe (n 206), Article 17.

[9] UNGA, ‘Seizing the opportunities of safe, secure and trustworthy artificial intelligence systems for sustainable development’ Seventy-eighth session Res 78/265 (21 March 2024).

[10] Ibid, preambular para 2

[11] Ibid, para 6.

[12] Ibid, para 3.

[13] Ibid, para 9.

[14] Baker McKenzie, ‘International: The United Nations adopts its first resolution on AI’ <insightplus.bakermckenzie.com/bm/data-technology/international-the-united-nations-adopts-its-first-resolution-on-ai#:~:text=On%2021%20March%202024%2C%20the,intelligence%20systems%20for%20sustainable%20development%22>.

[15] Baker McKenzie (n 339) para 6 (a).

[16] Ibid.