Generative AI (GenAI) systems have quickly gone mainstream with tools such as ChatGPT, Bard, GitHub Copilot and AlphaCode used by millions of people every day. Companies are adopting the technology to create new business opportunities, and start-ups are competing for venture capital.

GenAI applications have already begun to yield benefits in areas including code development, creative industries and arts, education, healthcare, and Internet search. More broadly, GenAI is expected to drive productivity gains and promote innovation and entrepreneurship.

GenAI has a role to play in addressing pressing societal challenges. In healthcare, GenAI can generate medical images to enhance limited datasets and create training samples and validate machine learning models. In pharmaceutical research, it can generate and optimise molecular structures with desirable properties for new drugs. As a tool to address the climate crisis, GenAI can analyse imagery and data about flooding and crops to help people make better decisions. These are just some ways GenAI is supporting progress towards achieving the UN Sustainable Development Goals (SDGs).

Policymakers must consider the implications of GenAI, such as potential impacts on labour markets and on-going debates around whether training generative AI systems on copyrighted material could constitute copyright infringement. Additionally, GenAI can perpetuate the biases inherent in training datasets. It can be used to generate and disseminate disinformation, deepfakes, and other manipulated content.

These campaigns can lead to broader negative impacts that can affect people, governance and the planet. For example, social, political, and economic repercussions could include disinformation on key scientific issues, perpetuating stereotypes and discrimination, distorting public discourse to create and spread conspiracy theories, influence elections, and distort markets.

In 2023, international governance of GenAI began to take shape. In updating the Guidelines for Multinational Enterprises on Responsible Business Conduct, the OECD set out expectations on the role of businesses in identifying and addressing AI-related harms. The G7’s reviewed generative AI’s opportunities and challenges and agreed to promote safety and trust as in areas such as governance, intellectual property rights, disinformation and responsible use of GenAI, resulting in the Hiroshima Process on GenAI.

The United Kingdom convened the first international AI Safety Summit. The European Union provisionally agreed to a final text for the EU AI Act, the first legal framework for developing and deploying AI systems, and the United Nations AI Advisory Body met for the first time and issued a report. National efforts, such as an AI Executive Order in the United States, have also taken shape. 

Rapid developments continue into 2024, with the OECD updating the OECD AI Principles, the Council of Europe poised to finalise the Convention on AI, and the UN set to launch a Global Digital Compact. At the national level, the United Kingdom and the United States are building AI Safety Institutes that focus on issues related to GenAI and more, and potential new policies and governance frameworks are on the horizon, such as in Brazil, Canada and Singapore.