Generative AI can create text, video, and images, and solve advanced tasks. As a result, it has become cheaper and easier to falsify content, spread disinformation, mislead, and manipulate.
Halfway through the super election year of 2024, where half of the world’s population is heading to the polls, it is clear that new AI tools are being used in election campaigns. Just as social media once transformed political campaigning, elections are now in the process of being reshaped again.
Examples of the use of generative artificial intelligence in election campaigns:
- USA: During the New Hampshire primary in January, voters received phone calls from a fake AI-generated imitation of President Joe Biden, urging them not to vote.
- Pakistan: Former Prime Minister Imran Khan used generative AI to spread political messages from his prison cell. The ex-prime minister wrote down his message, and an AI-generated clone of his voice was used to disseminate the speech as an audio clip and video.
- India: During India's election campaign, millions of deepfakes circulated, i.e., convincing, fake, and increasingly AI-generated images, videos, and audio clips. Political parties "cloned" their politicians and used them to reach more voters with personalized messages. Real-time translation of political speeches was also tested in the Indian election campaign – and is considered a great success.
- Indonesia: AI-generated images, videos, and audio clips of the deceased dictator Suharto were spread during the campaign. Several parties have "resurrected" well-known politicians to deliver today's campaign messages to voters.
Widespread use, high risk
It is too early to say whether the use of generative artificial intelligence in election campaigns has had a positive or negative effect on trust in politicians, political parties, and democracy itself. AI-generated propaganda, parody, and political satire flourished in the Indian election campaign, but the use of such tools in elections is not new. Nor is it considered illegal in democratic countries. Despite the extensive use of new AI tools in campaigns, the feared deepfake doomsday appears to be delayed.
What is certain, however, is that the technology is being used, its applications are growing, and limitations are few. This makes its use unpredictable and opaque, complicating efforts to map, manage, and mitigate risks.
The Norwegian Board of Technology’s project
The NBT will monitor the 2024 election year and how generative artificial intelligence is used in connection with election campaigns and voting. At the same time, we will highlight political solutions and options for Norway ahead of the 2025 parliamentary election.
The project will consider the following questions:
- How is generative artificial intelligence being used in the super election year of 2024?
- What consequences might this have for the Norwegian parliamentary election in 2025?
- What rules are needed to ensure the safe and responsible use of AI tools in elections?
