Nearly half of coronavirus-related tweets are sent by bots, research finds

Twitter bots are spreading misinformation and fake news about the coronavirus as rapidly as the pandemic has infected the world, U.S. researchers have found. 

Kathleen Carley, professor of computer science at Carnegie Mellon University in Pennsylvania, said that out of the more than 200 million coronavirus-related tweets her team analyzed, 45 per cent were sent by bots.

"It's definitely not what we expected," Carley told As It Happens host Carol Off.

Typically, she says about five to 25 per cent of tweets are posted by bots and the study's results are "way higher than we ever thought to see."  

The research found that of the top 50 influential retweeters, 82 per cent are bots. And out of the top 1,000 retweeters, 62 per cent are bots.

Bots are small, automated computer programs that are able to perform a number of simple functions over the internet. Carley says they can engage in conversations, repeat back common phrases, or rebroadcast the same message over and over.

"All they're doing is retweeting messages by other groups or by other individuals," she said.

"For example, around the state media accounts for Russia and China, 95 per cent of the top three tweeters are bots. They're just retweeting those accounts."

Carley said the majority of messages that were being retweeted from these accounts were about what these governments were doing with respect to handling COVID-19, the disease caused by the coronavirus. 

U.S.-based bots flooding Twitter

While a large amount of the activity was coming from these two governments, a surprising surge in U.S.-based, or "homegrown," bots has also been flooding the platform, especially as the debate around reopening the American economy was heating up, she said. 

"There's conservative bots who are arguing against liberals. There are liberal bots arguing against conservatives. It's across the spectrum."

They're often part of the support mechanism for spreading disinformation, but they're not the originators of it.- Kathleen Carley

Other types of messages these bots were pushing included calling on people to donate money to certain organizations, or just simply introducing certain people to each other by mentioning them, she said. 

However, disinformation around coronavirus was not being generated by the bots, but instead was being created on different websites and blogs, then being tweeted out by bots.

"They're often part of the support mechanism for spreading disinformation, but they're not the originators of it," Carley said. 

"Not only are 45 per cent of the actors bots, [but] they're in very strategic positions in the network so they're actually often responsible for things that end up trending," she said. 

'Very ripe' for exploitation

The team used a combination of "artificial intelligence and social networks" to analyze different factors about each of the messages. These included who the user was, what their online behaviour was like and how the tweet was written, she said. 

Carley says mismatches between the timing of multiple tweets sent by the same actor, the location of the messages, as well as the content, language and use of typos all gave some clues as to if it had been posted by a real person or a piece of computer script.    

Carley says the pandemic is "very ripe" for exploitation because the crisis is global, and "information is spreading from one country to another, which means that it's a perfect opportunity for different nation states to position themselves using these bots to [spread] their messages." 

The pandemic is also affecting so many people and causing so much desperation that it's created the right conditions to conduct fraud by using bots, she said.  

"All of those factors combined to make this a very good nexus."

Written by Adam Jacobson. Interview produced by Chloe Shantz-Hilkes.