As AI makes Arctic languages accessible, fears of disinformation rise
The Greenlandic public debate used to be immune to foreign influence but in recent months it has become available to hostile foreign actors. A new study from the University of Greenland looked into the new threats.
In 2019, Ane Lone Bagger, then Greenland’s minister of foreign affairs, sent an official letter to Tom Cotton, a member of the US Senate. In the letter, bearing with the Government of Greenland’s official letterhead, she asked for American help to fund a Greenlandic referendum on independence.
Or so it was thought. But the letter turned out to be a fake, most likely the result of deliberate disinformation from a foreign actor. The main suspects, according to experts, were either Russia or China.
Since then, no examples of similar attempts at disinformation are known to have occurred. But the threat remains high, not least because new AI translation tools have made the Greenlandic population more accessible and, thus, more vulnerable.
“Just within the past few months, it has suddenly become possible for large language models like ChatGpt to produce high-quality translations of Greenlandic texts. This is completely new, and didn’t used to be the case with Google Translate, for instance,” Signe Ravn-Højgaard, Director of the Think Tank Digital Infrastructure, told Polar Journal AG.
Last week, with co-authors from the University of Greenland, she published a report on the threat of disinformation in Greenland, a report that was prompted by the new technology.
“Because of AI, spreading false information is now much cheaper so you can produce much more of it. You can also create images and automate its spread to reach more people. This raises the risk of disinformation and that’s why we’ve decided to investigate this now,” she said.
Threats across the Arctic
Disinformation is defined as deliberately false information spread with hostile intentions. This is different from ‘misinformation’, which is defined as false information shared without an intent to deceive. In the report, the researchers looked only at disinformation spread by foreign states.
Thankfully, the report concludes that, at this point in time, no disinformation has occurred in the Greenlandic Facebook groups and pages that were surveyed. Instead, according to Signe Ravn-Højgaard, the report should be seen as a baseline for future studies.
In the future, the risk of disinformation will only increase, and, according to another co-author of the report, Rasmus Leander Nielsen, the possible culprits are many:
“Typically, one would primarily look for influence from Russia or China. But in principle, it could also come from the USA, for example, a right-wing group or a think tank, Canadians with vested interests in mining or infrastructure, or some completely different party, Rasmus Leander Nielsen, head of the University of Greenland’s Nasiffik – Centre for Foreign & Security Policy, told Polar Journal AG.
“We did not find disinformation in this analysis, but if it had come from, for example, Turkey, Italy, or Iran, we would have analyzed it on equal terms as if it was from Russia,” he said.
The report finds that the increased threat is likely not unique to Greenland but relevant across the Arctic regions where AI is also revolutionizing the accessibility of their small languages.
As examples, Rasmus Leander Nielsen mentioned discussions with colleagues from Nunavut about increased threats of disinformation in Inuktitut, the local Inuit language. And from discussions with representatives of the Sami community in northern Scandinavia, he reported similar worries.
“Typically, [the disinformation is spread] to create discord; internally in Greenland, within the Kingdom of Denmark, or in the Western alliance,” Rasmus Leander Nielsen said.
The Danish Realm is most sensitive
The new study looked only at Facebook which is used by about 80 percent of the population of Greenland and is the largest social media in the country by far.
On Facebook, it identified 64 different pages, each of them with more than 500 followers, and 43 groups, each of them with more than 1000 members. In all of them, the primary language was Greenlandic. From here, hundreds of thousands of comments and millions of interactions were investigated.
To check for foreign interference the researchers looked at three things. Firstly, the sharing of links from sources known to spread disinformation. Secondly, coordinated activity; posts or links that were spread unusually fast, a common sign of disinformation. And thirdly, a qualitative analysis of discussions on foreign policy topics.
The third step included a search for keywords that, in Greenland, are likely to be politically sensitive; words like ‘mining’, ‘the Danish realm’, and ‘NATO’.
None of these methods indicated that disinformation campaigns from foreign states were occurring.
“We didn’t find any signs of disinformation in Greenland at the moment,” Signe Ravn-Højgaard said.
“In addition to that finding, we also identified topics that are likely to be the subject of disinformation campaigns in the future, and here, Greenland’s relationship to Denmark seems to be the most vulnerable topic. That’s where the most debate and sensitive disagreements are. We know from disinformation studies elsewhere that disinformation campaigns target the topics that spark the most debate and the strongest feelings,” she said.
Disinformation is new in Greenlandic
Greenland is home to only 57,000 people. To influence the country’s public debate and, in turn, election results, not a lot of people would need to have their minds changed, at least compared to elsewhere.
However, as Signe Ravn-Højgaard points out, Greenland’s small population may work to its advantage.
“In a small country like Greenland people will have friends in common with most people. They will be able to see from their shared friends which family or which town another Facebook user is from. This makes it almost impossible for anonymous users, or Russian bots, to show up out of the blue, and start spreading discord at a large scale,” she said.
On the other hand, though, the new threats of disinformation may take some unsuspecting people in Greenland by surprise. This goes both for disinformation from foreign states and from opportunistic scam-artists.
“Those of us who speak Danish or English have gotten used to this threat. Years ago, we learned that emails from Nigerian princes were probably scams, and slowly we have developed a healthy skepticism to ‘phishing’ emails from strangers, even the really credible ones we receive these days.”
“In Greenlandic language, however, this is brand new. This hasn’t been possible before. So now the crooks of the world, both scammers and those who engage in disinformation campaigns, may increasingly turn to Greenland and the Arctic,” Signe Ravn-Højgaard said.
Ole Ellekrog, Polar Journal AG
More on the topic:
