What follows is a set of five things we are tracking in the Swedish campaign's digital communications. We want to be transparent about what this is and what it is not. These are observations from the outside, based on available research and reporting. We do not know yet whether any of them will move votes. Some may be noise dressed up as innovation. A few could genuinely change how campaigns communicate in Europe. We will know more in September. For now, these are worth watching.



1. Political SEO for AI, Not Google



This is possibly the most structurally novel thing happening in any European campaign right now, and it is worth slowing down to explain what it actually means.

A report published in April 2026 by Internetstiftelsen, the Swedish Internet Foundation, found that a quarter of young first-time voters are now using AI chatbots such as ChatGPT, Copilot, and Gemini to understand political parties' positions, compare policy proposals, and in some cases upload their own views to find the best match. One in ten Swedes overall uses AI tools to discuss politics. The generation born in the 2000s, Internetstiftelsen notes, tends to ask AI tools rather than Google. The search engine is not dead. But for a significant and growing slice of the electorate, the AI chatbot is the first point of contact with political information.

Swedish parties have noticed. The Green Party's digital team confirmed they are now producing web content structured specifically for AI retrieval, writing in clear declarative formats. The Liberal Party Secretary explicitly stated that "clear, fact-based, and accessible language" is now critical for visibility in AI-generated responses. The Sweden Democrats' press chief confirmed they are "continuously optimising" their site to ensure AI tools relay their policies accurately.

This is not traditional SEO. Traditional SEO optimises for search engine ranking. This is AI content strategy: structuring your own information so that when a voter asks a chatbot what your party believes, the answer your party would give is the answer the chatbot gives back. The parties that crack this first will effectively be present in every AI conversation about Swedish politics that a voter has between now and September 13. The parties that ignore it will be represented by whatever the model finds, or does not find.

Whether this actually moves votes, we genuinely do not know. AI chatbots are not neutral aggregators, and the interaction between a voter's query and the chatbot's response is not something any party can fully control. But the experiment is real, the investment is happening, and the logic is sound. Worth watching very closely.



2. TikTok Goes From Fringe to First-Time Voter Infrastructure



The numbers from Internetstiftelsen on TikTok are stark enough that they deserve to sit on their own. Ahead of the 2022 Swedish election, 23% of first-time voters consumed political content on TikTok. Ahead of the 2026 election, that figure has risen to 48%. Nearly half of Sweden's first-time voters are finding political content on the platform. Among all Swedes, the figure rose from 3% to 8%, which sounds modest until you realise that the 2022 baseline was essentially zero for practical campaign purposes.

The Sweden Democrats' TikTok account has 197,600 followers and 11.6 million likes, making it one of the most followed political party accounts on the Swedish platform. Jimmie Åkesson, the SD party leader, has used the platform to deliver short, punchy direct-to-camera messages with the same format discipline that worked for him on traditional media: calm, direct, security-focused. The Social Democrats and Moderates have TikTok presences too, but the right's comfort with the format, possibly a reflection of lessons learned from international sister parties in the populist tradition, has given them an early advantage in a space that is now strategically important.

What makes Sweden's TikTok moment particularly interesting is the platform context. Meta banned paid political advertising. TikTok bans it entirely. This means that TikTok reach for political parties is almost entirely organic, driven by content quality, algorithmic favour, and authentic engagement. There is no option to buy your way to an audience. You earn it or you do not.

Does TikTok reach translate into votes? That is the question Swedish campaigns are running a live experiment on. We know it reaches first-time voters at scale. We know those voters are, according to Internetstiftelsen, particularly influenced by political discussions they find online. We do not know yet whether a TikTok follower at 18 is a reliable vote at the ballot box. Sweden will give us one of the cleanest tests of that relationship we have seen in European politics.



3. Podcast as Policy Channel, Not Just Personality Platform



The podcast shift in Sweden is real and accelerating. Political content consumption via podcasts rose from 17% to 20% of the general population between 2022 and 2026. Among first-time voters, it went from 17% to 31%. Among young voters broadly, 30% now consume political content via podcast.

What is interesting is not just the numbers but the format being used. The most striking example we have found is AI Sweden Podcast, which has been running a dedicated pre-election series inviting spokespeople from each of the eight parliamentary parties to discuss AI policy, regulation, sovereignty, and the labour market. By July 2026 it had featured representatives from the Centre Party, Moderates, Greens, Social Democrats, Liberals, Christian Democrats, and the Left Party. These are not soft interviews. They are substantive policy conversations about one of the most technically complex issues in the election.

What the format does is create something political advertising cannot: long-form credibility. A party spokesperson who can hold a 40-minute conversation about AI governance with a knowledgeable interviewer is demonstrating competence in a way that a 30-second ad or an Instagram Reel simply cannot. For parties trying to reach educated, high-engagement voters who are already suspicious of political messaging, the podcast invitation is a different kind of communication.

The Swedish parties do not fully control this format. They cannot choose the questions. They cannot edit the answers. The interview might go somewhere uncomfortable. That is both the risk and the value: when it works, it is more persuasive than any controlled content, precisely because it is not controlled. Whether Swedish parties are using podcasts as a genuine communication channel or just as one more box to tick on a media checklist is something we will be watching carefully.



4. The AI Disinformation Awareness Play: Turning Defence into Offence



Here is something unusual. Ninety percent of Swedes, according to Internetstiftelsen's 2026 research, believe that AI will be used to deliberately mislead voters through fake images or texts in this election. That is not a fringe concern. It is a near-universal expectation.

What some Swedish parties and campaign-adjacent organisations are doing with that expectation is genuinely interesting. Rather than ignoring the issue or hoping it does not come up, they are front-running it. The governing parties in particular are using the public's existing anxiety about AI disinformation as a frame for their own communications, positioning clear, transparent, attributable content as a contrast to the expected flood of synthetic manipulation.

We are watching a specific hypothesis play out in Sweden: whether being the party that says "here is how we use AI, here is what is real, here is how to tell" can become a trust signal rather than just a defensive communication. Forskning.se and Swedish academics at Umeå University have been publishing research specifically on AI manipulation risks ahead of the election, research that campaign teams are clearly aware of and in some cases actively citing. When a party's communications team starts citing election security research in its own transparency disclosures, something structurally new is happening.

This could be genuinely effective. It could also be earnest framing that nobody outside the already-engaged political class notices. We do not know. But the fact that Swedish parties are even attempting to turn AI transparency into a political asset is a notable departure from every other campaign we have watched this cycle.



5. The Threshold Threshold: Digital Targeting Around the 4% Line



This one is specific to the Swedish system but the principle is transferable. The Riksdag's 4% electoral threshold creates a structural crisis for small parties that is, uniquely, a digital communications problem as much as a political one. The Liberals are currently polling at 2.3%, well below the threshold. The Greens are at 6.8% but have fluctuated close to the line in recent cycles. A party that falls below 4% disappears entirely from parliament, taking its MPs, its staffers, its resources, and its agenda with it.

The digital targeting challenge this creates is unlike anything you face in a first-past-the-post or open-list system without a threshold. A party hovering near 4% needs to do two things simultaneously and digitally: retain its existing voters, who are at risk of defecting to safer options, and convince tactical voters elsewhere that saving a small party is worth the cost of their vote. That requires entirely different messaging, different targeting, different emotional registers, and different platforms.

What we are watching is whether any of the threshold-endangered parties have genuinely built a digital operation around this specific problem. Not general campaigning. Not leader visibility. A targeted, data-driven effort to find and hold the specific voters they need to clear 4%, identify which ones are wavering, and serve them content designed specifically for that persuasion moment.

The equivalent would be a fire suppression system rather than a general alarm. Most small parties run the general alarm. The few that build the suppression system tend to survive the threshold. Whether any Swedish party has built it digitally this cycle, we genuinely do not know yet. But the incentives to do so are clear, the tools exist, and September 13 will be a fairly clean test of who figured it out.



A Final Note on Watching, Not Predicting



Sweden goes to the polls in 67 days. The campaigns are live, the data is incomplete, and any claim to know which of these five experiments will work would be dishonest. What we are confident in is that Sweden's combination of digital literacy, civic engagement, a well-resourced research community studying the campaigns in real time, and a multi-party system that creates distinct strategic problems for different actors, makes it one of the most instructive elections in Europe to follow closely.

We will be watching. And we will report back.