A team of researchers at the University of Adelaide have found that as many as 80 percent of tweets about the 2022 Russia-Ukraine invasion in its early weeks were part of a covert propaganda campaign originating from automated fake ‘bot’ accounts.
An anti-Russia propaganda campaign originating from a ‘bot army’ of fake automated Twitter accounts flooded the internet at the start of the war. The research shows of the more than 5-million tweets studied, 90.2 percent of all tweets (both bot and non-bot) came from accounts that were pro-Ukraine, with fewer than 7 percent of the accounts being classed as pro-Russian.
The university researchers also found these automated tweets had been purposely used to drive up fear amongst people targeted by them, boosting a high level of statistically measurable ‘angst’ in the online discourse.
The research team analysed a massively unprecedented 5,203,746 tweets, sent with key hashtags, in the first two weeks of the Russian invasion of Ukraine from 24 February this year. The researchers considered predominately English-language accounts, with a calculated 1.8-million unique Twitter accounts in the dataset posting at least one English-language tweet.
The results were published in August in a research paper, titled “#IStandWithPutin versus #IStandWithUkraine: The interaction of bots and humans in discussion of the Russia/Ukraine war“, by the University of Adelaide’s School of Mathematical Science.
The size of the sample under study, of over 5-million tweets, dwarfs other recent studies of covert propaganda in social media surrounding the Ukraine war.
The little-reported Stanford University/Graphika research on Western disinformation, analysed by Declassified Australia in September, examined just under 300,000 tweets from 146 Twitter accounts. The Meta/Facebook research on Russian disinformation reported widely by mainstream media, including the ABC a fortnight later, looked at 1,600 Facebook accounts.
This ground-breaking study, exposing a massive anti-Russia social media disinformation campaign, has been effectively ignored by the mainstream Western establishment media. It’s become almost routine during the Russia-Ukraine war.
The disinformation blitz krieg
The Adelaide University researchers unearthed a massive organised pro-Ukraine influence operation underway from the early stages of the conflict. Overall the study found automated ‘bot’ accounts to be the source of between 60 to 80 percent of all tweets in the dataset.
The published data shows that in the first week of the Ukraine-Russia war there was a huge mass of pro-Ukrainian hashtag bot activity. Approximately 3.5 million tweets using the hashtag #IStandWithUkraine were sent by bots in that first week.
In fact, it was like someone had flicked a switch, when at the start of the war on 24 February, pro-Ukraine bot activity suddenly burst into life. In that first day of the war the #IStandWithUkraine hashtag was used in as many as 38,000 tweets each hour, rising to 50,000 tweets an hour by day three of the war.
By comparison, the data shows that in the first week there was an almost total absence of pro-Russian bot activity using the key hashtags. During that first week of the invasion, pro-Russian bots were sending off tweets using the #IStandWithPutin or #IStandWithRussia hashtags at a rate of only several hundred per hour.
Given the apparent long-range planning for the invasion of Ukraine, cyber experts expressed surprise that Russian cyber and internet responses were so laggard. A researcher at the Centre for Security Studies in Switzerland, said: ‘The [pro-Russian] cyber operations we have seen do not show long preparation, and instead look rather haphazard.’
After being apparently left flatfooted, the #IStandWithPutin hashtag mainly from automated bots, eventually fired up a week after the start of the war. That hashtag commenced appearing in higher numbers on 2 March, day 7 of the war. It reached 10,000 tweets per hour just twice over the next two days, still way behind the pro-Ukraine tweeting activity.
The #IStandWithRussia hashtag use was even smaller, reaching only 4,000 tweets per hour. After just two days of operation, the pro-Russian hashtag activity had dropped away almost completely. The study’s researchers noted the automated bot accounts ‘likely used by Russian authorities’, were ‘removed likely by pro-Ukrainian authorities’.
The reaction against these pro-Russian accounts had been swift. On March 5, after the #IStandWithPutin hashtag had trended on Twitter, the company announced it had banned over 100 accounts using the hashtag for violating its ‘platform manipulation and spam policy’ and participating in ‘coordinated inauthentic behaviour’.
Later that month, the Ukraine Security Service (SBU) reportedly raided five ‘bot farms’ operating inside the country. The Russia-linked bot operators were reportedly operating through 100,000 fake social media accounts spreading disinformation that was ‘intended to inspire panic among Ukrainian masses’.
Unfiltered and independent research
The landmark Adelaide University research differs from these earlier revelations in another most unique and spectacular way.
While the Stanford-Graphika and Meta research was produced by researchers who have long-term deep ties to the U.S. national security state, the Adelaide University researchers are remarkably independent. The academic team is from the university’s School of Mathematical Science. Using mathematical calculations, they set out to predict and model people’s psychological traits based on their digital footprint.
Unlike the datasets selected and provided for the Stanford/Graphika and the Meta research, the data the Adelaide University team accessed didn’t come from accounts after they’ve been detected for breaching guidelines and shut down by Meta or Twitter.
Joshua Watt is one of the lead researchers on the university team, and is a MPhil candidate in Applied Mathematics and Statistics from the university’s School of Mathematical Sciences.
He told Declassified Australia that the dataset of 5-million tweets was accessed directly by the team from Twitter accounts on the internet using an academic license giving access to the Twitter API. The ‘Application Programming Interface’ is a data communication software tool that allows researchers to directly retrieve and analyse Twitter data.
The fake tweets and automated bot accounts had not been detected and removed by Twitter before being analysed by the researchers, although some were possibly removed in the March sweep by Twitter. Watt told Declassified Australia that in fact many of the bot accounts behind the 5-million tweets studied are likely to be still up and running.
Declassified Australia contacted Twitter to ask what action they may have taken to remove the fake bot accounts identified in the University of Adelaide research. They had not responded by the time of going to press.
Critical tool in information warfare
This new research paper confirms mounting fears that social media has covertly become what the researchers call ‘a critical tool in information warfare playing a large role in the Russian invasion of Ukraine’.
The Adelaide University researchers tried their best to be noncommittal in describing the activities of the fake Twitter accounts, although they had found the vast majority—over 90 percent—were anti-Russian messages. They stated: ‘Both sides in the Ukrainian conflict use the online information environment to influence geopolitical dynamics and sway public opinion.’
They found the two main participating sides in the propaganda war have their own particular goals and style. ‘Russian social media pushes narratives around their motivation, and Ukrainian social media aims to foster and maintain external support from Western countries, as well as promote their military efforts while undermining the perception of the Russian military.’
While the research findings concentrated on automated Twitter bots, there were also findings on the use of hashtags by non-bot tweeters. They found significant information flows from non-bot pro-Russian accounts, but no significant flows from non-bot pro-Ukraine accounts.
As well as being far more active, the pro-Ukraine side was found to be far more advanced in its use of automated bots. The pro-Ukrainian side used more ‘astroturf bots’ than the pro-Russians. Astroturf bots are hyper-active political bots that continuously follow many other accounts to increase follower count of that account.
Social media role in boosting fear
Crucially, the University of Adelaide researchers also investigated the psychological influence the fake automated bot accounts had on the online conversation during those early weeks of the war.
These conversations in a target audience may develop over time into support or opposition towards governments and policies—but they may also have more instant effects influencing the target audiences’ immediate decisions.
The study found that it was the tweets from the fake ‘bot’ accounts that most drove ‘an increase in conversations surrounding angst’ amongst people targeted by them. They found these automated bot accounts increased ‘the use of words in the angst category which contains words related to fear and worry, such as “shame”, “terrorist”, “threat”, “panic”.’
By combining the ‘angst’ messaging with messages about ‘motion’ and geographical locations, the researchers found ‘the bot accounts are influencing more discussion surrounding moving/fleeing/going or staying’. The researchers believe this effect may well have been to influence Ukrainians even away from the conflict zones to flee from their homes.
The research shows that fake automated social media ‘bot’ accounts do manipulate public opinion by shaping the discourse, sometimes in very specific ways. The results provide a chilling indication of the very real malign effects that mass social media disinformation campaigns can have on an innocent civilian population.
Origins of the Twitter bot accounts
The researchers report that the overwhelming level of Twitter disinformation that was anti-Russian was from bots ‘likely [organised] by pro-Ukrainian authorities’.
The researchers asserted no further findings about the origin of the 5-million tweets, but did find that some bots ‘are pushing campaigns specific to certain countries [unnamed], and hence sharing content aligned with those timezones’. The data does show that the peak time for a selection of pro-Ukrainian bot activity corresponded with being between 6pm and 9pm across U.S. timezones.
Some indication of the origin and the target of the messages, could be deduced from the specific languages used in the 5-million tweets. Over 3.5-million tweets, or 67 percent, were in the English language, with fewer that 2 percent in Russian and Ukrainian.
In May 2022, National Security Agency (NSA) Director and U.S. Cyber Command chief, General Paul Nakasone, revealed that the Cyber Command had been conducting offensive Information Operations in support of Ukraine.
‘We’ve conducted a series of operations across the full spectrum: offensive, defensive, [and] information operations,” Nakasone said.
Nakasone said the U.S. has been conducting operations aimed at dismantling Russian propaganda. He said the operations were lawful, conducted through policy determined by the U.S. Defense Department and with civilian oversight. Nakasone said the U.S. seeks to tell the truth when conducting an Information Operation, unlike Russia.
U.S. Cyber Command had deployed to Ukraine a ‘hunt forward’ cyber team in December to help shore up Ukraine’s cyber defences and networks against active threats in anticipation of the invasion. A newly formed European Union cyber rapid response team consisting of 12 experts joined the Cyber Command team to look for active cyber threats inside Ukrainian networks and to strengthen the country’s cyber defences.
The U.S. has invested $40 million since 2017 in helping Ukraine buttress its information technology sector. According to U.S. Deputy Secretary of State Wendy Sherman, the investments have helped Ukrainians ‘keep their internet on and information flowing, even in the midst of a brutal Russian invasion’.
Wars and lies in our pockets
With the rise of the internet, war and armed conflict will never be the same again. Others have noted that the Russian invasion of Ukraine has ushered in ‘a new digital era of military, political and economic conflict’ being manipulated by ‘laptop generals and bot armies’.
‘In all dimensions of this conflict, digital technology plays a key role—as a tool for cyberattacks and digital protest, and as an accelerator for flows of information and disinformation,’ wrote one.
‘Propaganda has been a part of war since the beginning of history, but never before could it be so widely spread beyond an actual conflict area and targeted to so many different audiences.’
Joshua Watt, one of the lead researchers on the University of Adelaide team that conducted the landmark study, summed it up. ‘In the past, wars have been primarily fought physically, with armies, air force and navy operations being the primary forms of combat. However, social media has created a new environment where public opinion can be manipulated at a very large scale.’
‘CNN brought once-distant wars into our living rooms,’ another stated, ‘but TikTok and YouTube and Twitter have put them in our pockets.’
We are all carrying around with us a powerful source of information and news media—and also, most certainly, disinformation that’s coming relentlessly at us from influence operations run by ‘bad actors’ whose aim is to deceive.
Peter Cronau is an award-winning investigative journalist, writer, and film-maker. His documentaries have appeared on ABC TV’s Four Corners and Radio National’s Background Briefing. He is an editor and cofounder of DECLASSIFIED AUSTRALIA. He is co-editor of the recent book A Secret Australia—Revealed by the WikiLeaks Exposés.