Iceland Votes Under Shadow of AI-Amplified Disinformation

Icelanders are voting today, August 29, in a referendum that will determine whether the government should resume EU accession negotiations, frozen for years. The vote does not yet decide membership itself, but in the weeks leading up to it, the country's president, police and media literacy organizations publicly warned of a growing wave of disinformation, including AI-generated content.

A referendum dividing the island

Iceland has held EU candidate status for years, but accession negotiations were frozen in 2015. The current government, formed after elections with a promise to hold a referendum on the issue, put the question directly: whether to return to talks with Brussels. Commentators have compared the atmosphere around the vote to Britain's Brexit referendum, pointing to similar campaign intensity and social divisions.

Behind the membership debate, however, a separate issue has emerged: the quality of the information Icelanders are basing their decision on. The country of roughly 400,000 people was long considered unusually resistant to disinformation, thanks in part to its language barrier, geographic isolation and high level of education.

The president's warning

Icelandic President Halla Tómasdóttir warned of the risks tied to AI-generated content, noting that AI can produce material that looks credible but is misleading and spreads rapidly online. Iceland's foreign minister raised similar concerns, linking disinformation, artificial intelligence and possible foreign interference in a single warning ahead of the vote.

Netvís, a media literacy organization that works with police and social platforms to remove harmful content, launched a public portal ahead of the referendum for reporting suspected attempts to manipulate public opinion. Its representative expects the number of reports to rise as voting day approaches.

We know that as things heat up, we expect to receive such reports - Stefanía Ragnarsdóttir, Netvís

However excellent a tool AI can be, it is not a truth machine - warning cited by Netvís

Troubling numbers

Research cited by Poynter shows the scale of the shift. 71 percent of Icelanders said they had encountered fake news online in the past year, up from about 59 percent in 2022. Roughly a quarter of respondents admitted forming a mistaken opinion about politicians or public figures based on misleading information.

Experts also point to the import of foreign narratives into Iceland's public debate. Nearly 40 percent of those surveyed believe in the existence of a so-called deep state, even though the concept previously had no domestic history of discussion in Iceland and was, as one account put it, adapted from foreign discourse. A similar rise was recorded in claims that migration from the Middle East is deliberately undermining Nordic and Christian values, from 23 to nearly 40 percent in four years.

A fading resilience

For years, three factors protected Iceland: a small population, in practice too small to make a mass disinformation campaign worthwhile, a language barrier limiting the inflow of foreign-language content, and a wealthy, highly educated population. The same research shows, however, that this protection is weakening amid widespread internet access, social media use and near-universal English proficiency, which opens Icelanders to content produced abroad, including machine-generated material.

Iceland's National Police Commissioner confirmed that authorities are monitoring public debate and social media for possible interference by foreign actors, checking for fake accounts, sites with unclear ownership structures, and potential foreign funding sources for referendum campaigns. Police noted, however, that they have not yet disclosed evidence that any foreign state is behind such activity.

For Polish readers, the Icelandic case is a reminder that AI-amplified disinformation around referendums and elections is not confined to large countries covered by the EU's AI Act. Even a small democratic society long considered resilient can, within a few years, face the same set of threats previously observed in larger votes across Europe and the United States, and the response mechanisms, a public reporting portal, police monitoring, warnings from top officials, look similar regardless of a country's size.