The UN General Assembly Adopts U.S.-Led Resolution on Safe, Secure, and Trustworthy Artificial Intelligence

The UN General Assembly has adopted without a vote a U.S.-led resolution “to promote safe, secure and trustworthy artificial intelligence systems to accelerate progress towards the full realization of the 2030 Agenda for Sustainable Development.”¹ Co-sponsored by over 120 member states,² the resolution—the Assembly's first on artificial intelligence (AI)—calls for “inclusive and equitable access to the benefits of digital transformation and . . . artificial intelligence,” emphasizes the respect for human rights in the use of AI, and encourages the promotion of safe, secure, and trustworthy AI, including through domestic regulation.³ The resolution establishes goals and principles but does not create rules, establish a process for drafting a binding international instrument, or propose any form of international AI governance.⁴ Though there are many multilateral initiatives setting out principles concerning AI's advance, the resolution uniquely links sustainable development with AI in a comprehensive manner and with broad support. “This resolution was a massive step towards . . . building and deploying AI systems that save lives, end poverty, protect our planet, and create a safer, more equitable world,” Ambassador Linda Thomas-Greenfield, the U.S. Permanent Representative to the United Nations, said immediately following the General Assembly's decision.⁵ The resolution follows through on President Joseph R. Biden, Jr.'s direction in his recent Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence that the United States should “lead efforts to establish a strong international framework for managing the risks and harnessing the benefits of AI.”⁶

As Ambassador Thomas-Greenfield noted in her speech to the General Assembly,⁷ the resolution builds upon a dizzying array of international initiatives, developed across organizations and groups, that establish high-level, non-binding frameworks for the regulation of AI. Some of these date back three to five years and more, such as the Organisation for Economic Co-operation and Development's Principles on Artificial Intelligence,⁸ the G20 Principles for Responsible Stewardship of Trustworthy AI,⁹ and the UN Educational, Scientific and Cultural Organization's Recommendation on the Ethics of Artificial Intelligence.¹⁰ But many have been issued only within the past year, as the rapid pace of AI's development, particularly in generative AI, has propelled governments to establish common baselines for cooperation and control. In October 2023, G7 leaders welcomed the International Guiding Principles for Organizations Developing Advanced AI Systems and a voluntary Code of Conduct for AI developers.¹¹ Also in October, Chinese President Xi Jinping introduced the Global AI Governance Initiative at the third Belt and Road Forum for International Cooperation.¹² The following month, at the AI Safety Summit hosted by the United Kingdom, twenty-eight states and the European Union endorsed the Bletchley Declaration in which the participants agreed on the need to identify AI risk and build risk-based policies, including safety testing.¹³ And in December, the Global Partnership on Artificial Intelligence, a collection of states and the European Union that seeks to facilitate international collaboration in the development of AI, issued its 2023 Ministerial Declaration, which underscored the importance of “AI projects aimed at addressing societal problems and global challenges, maximising the benefits and mitigating associated risks.”¹⁴

The resolution's negotiation and adoption also come amid a surge in domestic and regional AI regulation. Last year, the Cyberspace Administration of China issued Measures for the Administration of Generative Artificial Intelligence Services based on “the principle of giving equal importance to development and security.”¹⁵ The law prohibits generative AI from “generat[ing] content . . . such as inciting subversion of state power, overthrowing the socialist system, endangering national security and interests, damaging the national image, inciting the secession of the country, [and] undermining national unity and social stability.”¹⁶ In May 2024, the European Council gave final approval to the Artificial Intelligence Act, a regulation that establishes requirements and obligations based on AI risk and bans certain AI systems and uses, such as social scoring and predictive policing.¹⁷ Also in May, the Council of Europe adopted the Framework Convention on Artificial Intelligence, Human Rights, Democracy and the Rule of Law, the first binding international agreement regulating AI.¹⁸ The Convention aims “to ensure that activities within the lifecycle of artificial intelligence systems are fully consistent with human rights, democracy and the rule of law” and requires parties to take measures accordingly.¹⁹ Consultations are ongoing at the African Union on the development of a Continental AI Strategy.²⁰

In the United States, President Biden issued the first comprehensive executive order on AI in October 2023.²¹ The order emphasizes the development of new standards, guidelines, programs, and other measures pertaining to: AI safety and security; critical infrastructure; consumers, patients, passengers, and students; privacy; equity and civil rights; and innovation and competition.²² It mandates that: developers of powerful AI systems share safety test results with the U.S. government;²³ federal agencies set standards and create tests to ensure AI systems are safe, secure, and trustworthy;²⁴ and agencies use their authorities to establish standards and best practices to protect against AI-enabled threats such as fraud and deceptive content.²⁵ Further, it seeks to advance American leadership abroad by working with other nations to ensure the safe, secure, and trustworthy deployment of AI globally.²⁶

The resolution builds upon these global and domestic initiatives and follows extensive consultations among member states, as well as with stakeholders from civil society and the private sector.²⁷ It goes beyond its predecessors in the extensive support it received and in meaningfully and comprehensively connecting sustainable development with AI and AI safety. Its twin premises are that “safe, secure and trustworthy artificial intelligence systems . . . [will] accelerate progress towards the full realization of the 2030 Agenda for Sustainable Development” and conversely that “improper or malicious design, development, deployment and use of artificial intelligence systems . . . pose risks that could hinder progress towards the achievement” of the 2030 Agenda and its Sustainable Development Goals.²⁸ The resolution therefore details the actions that are required to provide developing countries with access to digital technologies and AI (to “bridge the artificial intelligence and other digital divides between and within countries”) and the ways in which all countries should promote secure AI.²⁹

It first highlights the imperative of cooperation with developing countries to ensure equitable access to the benefits of AI technologies and to address existing digital divides.³⁰ The resolution recognizes the unique challenges faced by developing nations in keeping pace with rapid technological advancements and emphasizes the urgency of closing these divides.³¹ Specifically, it calls upon member states and other stakeholders to take action to expand participation in digital transformation and enhance digital infrastructure connectivity in developing countries, including through increased capacity building, partnerships, funding, and technology transfer.³² The resolution emphasizes that “human rights and fundamental freedoms must be respected, protected and promoted throughout the life cycle of artificial intelligence systems.”³³ It encourages the promotion of safe, secure, and trustworthy AI and provides a comprehensive list of measures to achieve this goal.³⁴ These include the need for the development and implementation of domestic regulatory and governance approaches that support responsible AI innovation and investment, as well as means to prevent and mitigate AI vulnerabilities and risks prior to AI's deployment and use.³⁵ It also notes the importance of fostering public understanding of the core functions, capabilities, limitations, and appropriate uses of AI systems, as well as promoting transparency, predictability, reliability, and understandability throughout their lifecycle.³⁶ It stresses the need to protect individuals from discrimination, bias, misuse, or other harms associated with AI systems.³⁷ It calls for measures to mitigate negative consequences for workforces, especially in developing countries, by fostering programs aimed at digital training, capacity building, supporting innovation, and enhancing access to the benefits of AI systems.³⁸ The resolution recognizes as well the importance of responsible and effective data governance and encourages greater international cooperation, collaboration, and assistance.³⁹

“Today,” Ambassador Thomas-Greenfield said upon the resolution's adoption, “all 193 members of the United Nations General Assembly have spoken in one voice, and together, chosen to govern artificial intelligence rather than let it govern us.”⁴⁰ Vice President Kamala Harris, who has led the administration's AI policy development, heralded the resolution's adoption as “a historic step toward establishing clear international norms.”⁴¹ Explaining the administration's approach to AI, she said that “AI must be in the public interest—it must be adopted and advanced in a way that protects everyone from potential harm and ensures everyone is able to enjoy its benefits. And we must also address the full spectrum of risk, from catastrophic risks to all of humanity to the harms felt by individuals and communities, such as bias.”⁴² Emphasizing the importance of AI and global development, she continued: “Too often, in past technological revolutions, the benefits have not been shared equitably, and the harms have been felt by a disproportionate few. This resolution establishes a path forward on AI where every country can both seize the promise and manage the risks of AI.”⁴³ She also pointed out that “the resolution makes clear that protecting human rights and fundamental freedoms must be central to the development and use of AI systems.”⁴⁴

The U.S. efforts in the General Assembly came at the same time that the United States is seeking support for the Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy, a non-binding text that sets out principles for the use of AI.⁴⁵ That instrument, which the United States drafted and proposed on its own, now has the backing of fifty-four countries.⁴⁶