The United Nations General Assembly passed its first resolution on AI on 21st March 2024 and passed with the support of all 193 member states. The text is available here to read.

For the purposes of this summary, I focused on key aspects of the UNGA Resolution with regard to future-facing safety measures from advanced AI systems and looked over points such as reduction in the global digital divide, inclusivity and disinformation risks.

5 key takeaways:

1. Calls for Adequate Governance Mechanisms:

• The resolution recognized the rapid acceleration in the AI landscape along with some encouragement for regulatory innovation, agile responses and the need for continued international dialogue.
• Perhaps my favourite line in the document was, “International approaches (have to) keep pace with the evolution of artificial intelligence systems and their uses.”

2.It had a Little Bit in it for Everyone:

The resolution:

• Called on businesses to adhere to the UN Guiding Principles on Business and Human Rights and implement the “Protect, Respect and Remedy” framework. It also alluded to the need for fair competition, particularly for SMEs in the space. At the same time, it called for appropriate safeguards to respect IP rights.
• It touched on user rights, safeguarding privacy, raising awareness of the use of AI systems and the negative consequences of AI on the workforce.
• For people concerned with future risks, there were risk assessments, evaluations, interoperable standards and the need for regulatory mechanisms to keep up with the evolution of AI systems.
• For people concerned with ethics and human rights, the resolution also recognized that improper use of AI may lead to risks to human rights.

3.Encouraged a Rights-Based User-Centric Approach:

• The resolution emphasized that human rights and international human rights law must be kept at the centre of discussions. This was a nice way to hint at AI systems that might not be aligned with human values.
• It also called for feedback mechanisms for users to report any vulnerabilities or anomalies, as well as provenance mechanisms.

4.Investment and Global Cooperation is the Way Forward:

• The resolution touched on the need for a shared international framework with interoperable technical tools, standards or practices while also respecting each country’s domestic regulatory needs.
• The resolution highlights the need for "increased collaboration" among public and private sectors, civil society, and academia to support regulatory frameworks in the face of AI risks.
• At various points, the need for investment in effective safeguards such as physical security and risk management systems was mentioned.

5.Presented a Lot of Potential Policy Levers:

• The resolution acknowledged the need to carry out evaluations, require disclosure of risk monitoring mechanisms, hinted at the need for harmonization of standards, logging AI misuse incidents and global information sharing to foster best practices.

Some Miscellaneous Thoughts:

• Given that there was a resolution focused solely on AI, it goes to show how important or disruptive leaders expect this tech to be.
• In the joint press release here, the sponsoring states acknowledge that AI “has enormous potential to shape our economies, societies” and that “As AI technologies rapidly develop, there is urgent need and unique opportunities for Member States to meet this critical moment with collective action.”
• The United Nations is a platform where agendas are generally kept high-level so that they can be adapted to each country’s needs(one must be mindful that international resolutions are non-binding as UNGA resolutions are soft law instruments).
• The resolution does set a valuable precedent for future legislation and research. It also took note of the UN Interim report by the High-Level Advisory Body on AI, looked forward to the development of a global digital compact and the overall review of the progress made since the World Summit on the Information Society. It acknowledges the unique role of the United Nations system in reaching a global consensus on safe, secure, and trustworthy AI in line with international law and the SDGs.
• The resolution kept emphasizing protection “throughout the life cycle of artificial intelligence systems.”
• The definitions of safe, secure and trustworthy AI felt overlooked. It makes sense that this phrasing was used by one of the key sponsors of this resolution, the US, who used the exact same wording in the executive order and the White House statement. There is still, for me, always a nagging feeling of “Safe according to which benchmarks?’ and “Do we even agree on what trustworthiness entails?”.  I don't think there will be objective answers to these questions, and my personal point of view is the UN is probably not going to give us those answers as different interpretations of the regulatory landscape evolve. 
  
  However, how these concepts are interpreted across different jurisdictions and by various stakeholders could highlight the challenges in achieving a consensus on AI governance.
• The resolution sets ambitious high-level goals. The practical challenges associated with its implementation deserve attention. Questions around how compliance can be monitored, the role of international organizations in supporting countries to meet these standards, and the potential for capacity-building initiatives could be explored.