South Africa was one of the key targets of a massive social media campaign mounted by Russia in support of its invasion of Ukraine – and it hosted many of the bots spreading Moscow’s propaganda, according to new research.
Evidence of the Twitter campaign emerges in an analysis of almost 350,000 tweets in support of Russia and its president, Vladimir Putin.
One in five of the tweets came from bots – profiles controlled by software – created after the invasion on 24 February, say German researchers.
Just over 10% of the bots involved in a “systematic, coordinated propaganda” campaign were based in South Africa, and the country also had one of the biggest retweet networks.
One stream of tweets using hashtags such as #iStandWithRussia and #iStandWithPutin said South Africans supported the Russian leader because of Moscow’s role in the struggle against apartheid.
Pro-Russian tweets peaked on 2 March, the day South Africa joined 34 other countries in abstaining when the UN general assembly overwhelmingly passed a resolution deploring the Ukraine invasion.
“Several of the countries with many pro-Russian messages also showed a pronounced role of bot activity,” the researchers say in a paper posted on the open-access platform arXiv.
India and the US each had 24% of the bots, SA 10.2% and Nigeria 7.9%. “The patterns remained robust across different methods for inferring geographic locations,” says the paper.
“Overall, countries that abstained from the UN vote had the highest relative frequency of bots (20.3%), in comparison to countries that voted against (14.9%) or approved (16.6%) the UN resolution.
“The accumulation of messages on the day of the vote gives rise to concerns that countries that abstained… were systematically targeted by Russian propaganda efforts.
“Strikingly, many bots that spread Russian propaganda had been created shortly before the UN vote, which indicates an intentional and planned manipulation of public opinion on Twitter.”
When they analysed the retweet network, which ensured the pro-Russia messages reached about 14.4-million people, the researchers found major clusters in South Africa and India.
“These countries exhibited relatively isolated retweet networks in which Russian propaganda was able to infiltrate the local online communities with little external influence,” they say.
“In comparison, accounts from the US did not show the same geographic clustering but were more broadly scattered over the retweet network. This suggests that there were differences in the coordination behind the propaganda campaign across countries.”
Bots were more active retweeters than human users but the researchers said both categories retweeted humans more than bots. “This is a crucial difference from earlier work on low-credibility content for which humans have been found to frequently retweet bots,” they say.
“[It implies] that it is difficult for bots to make Russian propaganda go viral among humans.”
However, “bots primarily aimed to expose users to organic, pro-Russian content from humans. By creating traffic around Russian propaganda, certain hashtags appeared as so-called ‘trending topics’ on the front page of Twitter and were thus visible to all users. This is especially alarming, since repeated exposure can lead people to perceive misinformation as accurate.”
The researchers, from universities in Munich and Giessen, said Russia’s use of propaganda to influence Western opinion – often to destabilise democracies by sowing doubt and polarising citizens – is not new.
But social media allows the manipulation of public opinion at an unprecedented scale, they say. “This is partially enabled by the low cost of producing high volumes of software-controlled social media profiles.
“Generally, bots are deployed to spread disinformation, fake news and hate speech on social media. In particular, they aid in the spread of low-credibility content.”
After identifying 349,455 propagandist tweets between 1 February and 31 July, the researchers used Botometer to classify the accounts where they appeared; 21% were bots, and they were responsible for 26% of retweets.