Almost four billion people in 70 countries will go to the polls this year. In Germany, there are several local and state elections and in June, EU citizens will vote for the European Parliament. With the increasing spread and accessibility of AI, the risk of voters being disinformed and influenced in their decision is growing. Experts from the European IT security manufacturer ESET assume that election manipulation by AI will occur in various forms.

Fake videos

“It is very likely that there will be attempts at manipulation by AI in elections this year, for example in the form of fake politician videos. Not only political parties in their own country have an interest in damaging their political opponents, but foreign actors will also try to intervene in the election campaign using AI,” explains Michael Klatte, IT security expert at ESET. “Voters in Germany must therefore be careful when consuming news on social media or other sources – AI technology is now so advanced that even professionals cannot recognise fake images and videos at first glance.”

Fake video of the Federal Chancellor

Generative AI can create very realistic deepfakes. The manipulated images and videos are intended to damage the image of politicians and stage events that never took place. At the end of last year, for example, a fake video of Olaf Scholzmade the rounds. As production technology becomes ever more sophisticated, it is becoming increasingly difficult to distinguish between authentic and fake content. To create such a video, all you need is the right raw data material, such as videos, images and audio recordings – all of which are available online en masse from politicians. AI tools that use this data to generate images, text and audio files are available online for a small fee and create authentic-looking media in no time at all.

Disinformation with AI in elections has long been a reality

In the USA, AI tools are already very popular for defaming political opponents. For example, an AI caller urged voters in the US state of New Hampshire not to take part in the primaries, thereby torpedoing the election campaign of incumbent US President Joe Biden. In Germany, disinformation campaigns via other channels are suspected: social media platforms and messenger services such as TikTok, Instagram, Telegram and WhatsApp in particular are already being used to spread misinformation – even if their operators are already doing a lot to counteract this.

Take a close look and check the source

TheDigitalServices Act, which came into force in the EU at the end of 2022, requires search engine and social media operators to mitigate risks in election processes. This also includes responding to disinformation. Deleting questionable content can still take some time. Users should therefore adhere to the following tips when they come across seemingly explosive news and scandals involving political dignitaries on the internet:

Check the source and look for signs of inconsistencies, inaccuracies or overly emotional content that could indicate possible manipulation. Although AI technologies are getting better and better, errors and inaccuracies still occur. For example, hands often look unnatural in images and lip movements in videos appear inaccurate or asynchronous.

The German government provides further tips online.