Study: Swedes Fear Election Interference – AI Highlighted as a New Threat

More than half of Swedes believe the country will be subject to foreign election interference ahead of the election. At the same time, the Swedish Psychological Defence Agency assesses that the Swedish election is not currently a prioritized target for foreign threat actors. However, the development of generative AI has changed the playing field and made it significantly easier to produce convincingly manipulated material.

According to a new survey commissioned by the Swedish Psychological Defence Agency (MPF), 64 percent of respondents believe it is very or quite likely that Sweden will face attempts at election interference from foreign powers or other foreign threat actors during the year.

Additionally, 56 percent believe that Sweden actually will be subjected to election interference. The survey includes 1,360 responses and was conducted during weeks 27–31.

Despite the widespread concern, the MPF currently sees no indications of any large-scale, coordinated foreign information campaign aimed at influencing the Swedish election. The agency also assesses that the situation could change rapidly.

– The election is taking place under uncertain and serious security conditions, but it is not currently considered a prioritized target for foreign threat actors, says Magnus Hjort, Director General of the MPF.

However, this does not mean the threat is nonexistent. According to the MPF, the main threat actors are often opportunistic. If a political conflict or debate presents an opportunity to create divisions or influence public opinion, it can be exploited even without a major, pre-planned campaign.

AI Lowers the Threshold

The major development ahead of this year’s election, according to the agency, is the rapid advancement of generative AI. The technology makes it possible to quickly produce large quantities of images, video clips, and other materials that can appear real.

Previously, producing that type of material required significantly greater technical resources. Now, even individual actors can create material credible enough to spread on social media.

MPF especially warns that AI-generated material does not need to be spread through large, coordinated bot networks to make an impact. A manipulated image or video can first be posted by a single user and then spread organically if it provokes strong emotions or fits into an ongoing political conflict.

There are also recent examples of AI already being used in influence attempts against Sweden. In early September, SVT reported on MPF’s mapping of the AI-driven botnet “Holiday.” Around 1,200 accounts were used in 2025 and 2026 for information influence campaigns targeting Sweden and other countries, with over 50,000 posts, comments, and shares.

Strong Resilience

At the same time, MPF points out a more fundamental risk — as it becomes harder to determine what is genuine, trust in real images, videos, and information may also begin to erode. Those who no longer know what can be trusted may become more susceptible to false information, but also more skeptical of accurate information.

For the Swedish election, the situation is thus twofold. There is currently no sign of any large-scale foreign campaign targeting the election, according to MPF. But technological developments have made it easier than ever to quickly create material that can be used to influence voters, increase divisions, or undermine confidence in the democratic system.

– The Swedish population’s resilience to information influence is strong, says Magnus Hjort.