Can artificial intelligence be used to undermine elections?

The UK and the US are both heading towards national elections. In the next couple of years almost every major democracy will also hold the most important national elections in their systems. Previous elections in the UK and the US and a referendum in the UK saw spikes in political disinformation online, as well as online exercises in profiling and targeting for the purpose of political influencing. Similar activities have been identified around elections worldwide.The online influencing that happens around these coming elections will not simply be repeats of what we have seen before. Since the last US and UK general elections, artificial intelligence has continued to develop fast. New applications including large language models have been developed, and some have become widely available. It is very likely that some of these new tools will be used in attempts to influence the functioning and the outcomes of those elections. Some of those uses may fall within the range of safe and legitimate political activity. Others may be harmful and even dangerous to democratic processes.Collectively, we should improve our understanding of what could go wrong and how we should be protected from new harms, before the elections take place.