Enlarge image

Voting in the primaries in the USA: Manipulation with fake calls

Photo: Sam Wolfe / REUTERS

Fake calls from politicians, fake photos or videos: leading technology providers want to prevent content created by artificial intelligence (AI) from affecting political elections around the world.

At the Munich Security Conference on Friday, representatives from 20 companies - including Adobe, Amazon, Google, IBM, Meta, Microsoft, OpenAI, TikTok and X (Twitter) - committed to working together to better detect and combat malicious AI content.

Tech providers are primarily targeting AI-generated audio, video and image files that deceptively mimic or alter the appearance or voice of political candidates, election workers and other key players in a democratic election. The initiative also targets digital content that provides voters with false information about when, where and how the elections will take place.

Especially in the USA, where the presidential election will take place on November 5th, there is already plenty of experience with manipulative AI content in the current election campaign. There, a telephone robot recently asked people with the vote of President Joe Biden not to cast their vote during the primary elections in New Hampshire. The voice generated by an AI can hardly be distinguished from the real voice of the US President.

The head of the Munich Security Conference, Christoph Heusgen, said elections are the beating heart of democracies. The agreement is a crucial step to promote the integrity of elections, increase societal resilience and create trustworthy tech practices.

Kent Walker, President of Global Affairs at Google, said: "We cannot allow digital misuse to threaten the ability of artificial intelligence to improve our economies, create new jobs, and advance advances in health and science." Microsoft- President Brad Smith said that as society reaps the benefits of AI, signatories to the agreement have a responsibility to ensure that these tools are not weaponized in elections.

As part of the agreement, the companies involved want, among other things, to re-evaluate their AI models to understand the risks they could pose in connection with fraudulent election content. At the same time, the companies want to support efforts to raise public awareness of this challenge. The signatories also want to promote media literacy and the resilience of society as a whole.