Sweden's institutional trust needs an evidential backstop

Sweden has long regarded trust as one of its national strengths, and international comparisons suggest that this reputation is well deserved. Trust helps make everyday life possible without requiring us to question every interaction, investigate every statement, or demand evidence of everyone's intentions. It is equally important to our relationships with public institutions and businesses, whose communications we routinely rely upon when making decisions. Yet the authority attached to a familiar and trusted name is increasingly being exploited by those who have never had the right to speak on its behalf.

During 2025, for example, residents across Sweden received fraudulent messages warning that their drinking water was contaminated, and this was often presented as ‘official’ communications from their municipality. According to an investigation by Sveriges Radio, some messages were apparently intended to persuade residents to let strangers into their homes, while the affected authorities were left explaining that no such warnings had been issued. Similar problems appeared when Riksbanken had to disown fabricated videos portraying Governor Erik Thedéen promoting investment opportunities that same year.

In July, the Swedish government published its first national strategy for psychological defence, recognising the importance of strengthening society's resilience against information influence. Among its priorities are reliable communication channels and ensuring that authoritative information reaches the public. The strategy addresses an important part of the problem, such as reminding citizens to verify information through existing resources such as Krisinformation.se, however, it leaves room for a further question about what happens to institutional communications once they begin circulating beyond those original channels.

Businesses also face similar problems when their own communications are reproduced and manipulated beyond their control. In late 2024, for example, Danish-Swedish dairy group Arla became the target of a wave of misinformation following its announcement of trials involving a methane-reducing cattle-feed additive. Altered screenshots of the company's original announcement and press release circulated online, accompanied by allegations that its milk was poisonous and calls for consumers to boycott its products. The controversy spread across Britain and Denmark, with consumers posting videos of themselves pouring away Arla milk, while the company was forced to respond publicly to claims that bore little resemblance to what it had originally communicated. This particular episode is a cautionary example of how an authentic corporate announcement can become the raw material for an entirely different narrative once it begins circulating beyond the publisher's control.

In an ideal world, citizens, journalists and businesses should question what they encounter, and that expectation is not unreasonable in today’s information landscape. However, their ability to check an authoritative claim depends partly on whether the organisation has made reliable evidence publically available. Searching a company's website and finding no corresponding statement may raise suspicions, although it does not necessarily establish that the statement was never made, but if that company has established a clear and verifiable record of its official communications, including where and how they are authorised and published, an announcement that cannot be traced to that record provides a rather more substantial reason for doubt.

This creates an interesting opportunity for organisations to establish not only what they have communicated, but also a clearer boundary around what can legitimately be attributed to them. A sufficiently comprehensive record of authenticated publications, together with clearly defined practices for issuing, correcting and withdrawing statements, would provide an independent point of comparison when suspicious communications appear. The absence of a matching record would not conclusively establish that something was fabricated, but it would offer evidence that the communication does not correspond to the organisation's established publication process. For businesses attempting to challenge fabricated statements, and for journalists deciding whether to repeat them, that additional evidence could be particularly valuable, particularly as these types of disinformation attacks tend to gather momentum very quickly.

Much of my academic work has concerned how narratives change as they move between people and technological systems, and there is a philosophical dimension to this problem that is easily overlooked, and worth mentioning. When a public authority issues a warning, or a company announces a decision, the communication is itself an ‘illocutionary’ act, performed by someone with the authority to do so. Its significance depends upon that act having taken place, rather than merely upon the words appearing somewhere. Yet, whilst our digital systems are remarkably good at preserving and reproducing words, they are considerably less consistent about preserving the evidence of who originally communicated them, when, and under what contextual circumstances.

Of course, the arrival of generative AI makes this distinction increasingly urgent. According to Internetstiftelsen, 71 per cent of Swedish white-collar workers now use AI at work, and these systems are already becoming intermediaries through which people encounter, interpret and act upon information. An AI assistant will reproduce a statement, paraphrase its meaning, or attribute it to an apparently credible source without necessarily establishing that the source published it originally. Equally, it could help perform that verification, provided an accessible and trustworthy record exists against which the claim can be checked.

European electronic trust services provide many of the necessary foundations for authenticating digital records, while initiatives such as C2PA are developing common standards for establishing the provenance of digital media. Listed companies also operate under established disclosure and record-keeping requirements, so there are governance strategies already working towards a verifiable future. What deserves further attention however, is how these existing capabilities can support a common approach to verifiable communications, allowing consequential institutional statements to remain independently checkable even when they are copied, quoted or encountered through other systems.

Such an ideal would not intend to make institutional statements true, and nor should it; an authentic communication can still be inaccurate or misleading, and establishing who published it should never be confused with establishing whether its contents are correct. Nevertheless, preserving evidence of what an institution actually communicated provides a more reliable basis for public scrutiny, and makes fabricated attributions easier to challenge.

As Sweden develops its information resilience, there is an opportunity to extend the principles already established in electronic trust services and digital provenance, to the wider communications of public institutions and private organisations. Government, industry representatives, and standards organisations should work towards common expectations for how consequential communications are recorded, how changes and withdrawals are documented, and how their authenticity can be independently established, even when the information has circulated beyond its original source. Such an approach would recognise that information resilience is not exclusively a responsibility of the state, and that businesses have an equally legitimate interest in being able to establish what they have, and have not, communicated.

In an information environment where a convincing imitation can travel just as far as the original, preserving the evidence of who actually said something seems a reasonable responsibility to place upon those whose words are expected to carry authority.