The promises of AI are vast. Harnessed effectively, it can transform sectors and contribute to global productivity and prosperity. However, progress is accelerating at a speed far outstripping democratic processes and controls. With this comes risks to society and democracy and potentially loss of human control over AI. In democracies, we are not yet set up to address these safety concerns.
The UK Safety Summit can be an important landmark in progressing towards global AI safety. While emphasising global cooperation on frontier AI safety, it is crucial to recognize the dangers and risks posed by existing AI technologies, especially when they challenge democratic systems. The Summit can help initiate a conversation across democracies on measures to address such risks. A proposed AI Safety Institute, or an expert monitoring group akin to the Intergovernmental Panel on Climate Change, can then provide essential support to democracies with impartial, reliable and timely assessments about AI progress and research on the impact of AI on society and democracy.
Institutional components of democracy will be key to mitigating harms and risks from AI. We need a revitalised global effort to support the development of robust, people-centred, trusted democratic institutions which can address the changes to democratic societies that AI will bring. This is long-term and essential work.