A new tool is tracking automated and otherwise questionable social media accounts as they sow disinformation, discord and division. Climate change is a target.

Silhouettes in front of a Twitter logo. Credit: Glenn Chapman/AFP/Getty Images

Scientists and others targeted by climate change-related Twitter attacks worry about the potential for online onslaughts to spread disinformation and manipulate public opinion. Credit: Glenn Chapman/AFP/Getty Images

CNN’s seven-hour climate change town hall for presidential candidates was not a TV ratings bonanza, but it set off a marked surge of activity on Twitter aimed at ridiculing the Democrats and dismissing the science.

“Climate change” became the top two-word trending topic on Twitter for several hours after the event among the accounts being tracked by Bot Sentinel, a free platform designed to track what it considers untrustworthy or automated accounts. It was quite an unusual feat for the topic to beat out—even temporarily—the phrase that sits almost constantly atop the trending list for accounts on Bot Sentinel’s watchlist: “President Trump.”

Scientists, activists and politicians who are engaged in climate policy say they are being besieged by a surge of online attacks. It is difficult to divine whether the bursts of “climate change”-related Twitter activity are spontaneous or part of coordinated campaigns; some experts say that likely a small number of influencers are touching off postings by a far larger number of followers. But in a post-2016 world that is keenly aware of the role that social media played in the election of Donald Trump, the targets of climate attacks are concerned about the potential for online onslaughts to manipulate opinion and neutralize growing public support for climate action.

“I believe this is a concerted effort, likely by bad state actors and fossil fuel interests, to create disinformation, discord and division as we approach the all-important UN Summit and children’s youth event later this month,” said climate scientist Michael Mann of Pennsylvania State University, a frequent target of attacks.

Twitter did not immediately respond to a request for comment on the activity.

Tracking ‘Trollbot’ Attacks

Bot Sentinel has become a tool for those trying to fight the spread of disinformation on Twitter. The platform uses an algorithm to identify accounts it labels as “trollbots”—those that frequently retweet known propaganda accounts, exhibit repetitive behavior or violate Twitter’s terms of service by harassing other users. Following CNN’s climate forum on Sept. 4, there was an unusually high 700 mentions of climate change in a 24-hour period from the 100,000-some accounts Bot Sentinel is tracking as trollbots.

When a topic like “climate change” trends among the trollbots, it is likely there is some amount of coordination involved, said software developer Christopher Bouzy, founder of the year-old Bot Sentinel platform.

“What we are noticing is these phrases are more than likely being pushed by accounts that have an agenda,” Bouzy said.

“It’s fascinating to see this stuff happen in real time,” he said. “Sometimes we can see literally five or 10 accounts able to manipulate a hashtag because they have so many people following them. It doesn’t take that many accounts to get something going.”

Bot Sentinel's Sept. 5 report for two-word key phrases. Credit: Bot Sentinel

Following the CNN climate forum, for example, many accounts began sharing video clip montages of the event edited together by Trump’s re-election campaign and by the far-right web outlet the Gateway Pundit to portray the candidates in the worst possible light.

“Democrat #Socialists want to ban: – everything made from plastic – red meat – nuclear power…I don’t think even Venezuelans have ever been this brainwashed!” commented @Condor_Law, an account rated “alarming” by Bot Sentinel—84 percent of its posts exhibit patterns that match the activity of trollbots (spreading information from sites known to be unreliable or violating Twitters terms of service and doing so repetitively.) “The Dems position is ban straws, portion our meat, take our guns, take our cars, abort on a massive scale for population control, and all so we can die from climate change in 11 years!!” another accounted rated “alarming,” @JenniferKrist28, chimed in.

Mann, the climate scientist, found himself on the receiving end of some of this activity after he posted a Tweet thanking CNN for a “full evening of informed, detailed climate change conversation.” “Climate change is a hoax,” replied @fdnymt, another account identified as exhibiting trollbot behavior by Bot Sentinel.

Swedish teen climate activist Greta Thunberg, who traveled to the United States for the UN Summit, has been a frequent target of toxic online attacks. Soon after Thunberg started her two-week sailboat voyage across the Atlantic, British political donor and co-founder of the Leave.EU campaign Arron Banks tweeted, “freak yachting accidents do happen in August.” Following the lead of figures like Banks, lesser-known Twitter handles have piled on.

Thunberg has brushed off the attacks: “When haters go after your looks and differences, it means they have nowhere left to go,” she tweeted. “I have Aspergers and that means I’m sometimes a bit different from the norm. And – given the right circumstances- being different is a superpower.”  U.S. climate change denier Steve Milloy, who served on Trump’s transition team, responded to the teenager on Twitter, “I don’t know about you… but I am not going communist because Greta the Climate Puppet believes she has ‘superpowers.'”

In Canada, Politicians and an Election Targeted

Government policies on climate change also appear to be triggering toxic online activity. Early this week, Canada Environment Minister Catherine McKenna, who is implementing that nation’s carbon tax policy, said that she has been assigned an additional security detail because of abuse she has received both online and in person.

And last week, a Canadian government agency issued a report concluding there was evidence of a campaign of coordinated, false social media postings around last April’s election for premier of Alberta—a campaign in which climate policy was a key issue. The majority of the questionable postings were in support of the eventual winner, conservative Jason Kenney, who won the election while pledging to repeal the province’s carbon tax, which he did in May, and to create a “war room” to combat the oil industry’s opponents.

The government agency, Rapid Response Mechanism Canada, was created as part of the G7 nation’s effort to respond to foreign influence in democratic elections. RRM Canada said there was no evidence of foreign actors involved in the questionable social media activity around the Alberta election. But, the agency said, “some national actors were disseminating material using tactics that were similar to those used by foreign actors.”

Gerald Kutney, an Ottawa-based consultant and author of a 2014 book on climate politics, said it has become difficult to discern what is genuine discussion and what is coordinated attack messaging on Twitter around climate change.

“It’s no longer social media, it’s bot media,” he said. “It’s difficult to know the difference.”

Twitter Deactivated a Way to Block ‘Trollbots’

Kutney endured so many online attacks after he got into a debate on climate change last fall with the Dilbert cartoon creator and prominent Trump supporter Scott Adams that he found a way to call for online help. He launched a hashtag, #climatebrawl, to signal when someone is under attack by climate deniers on Twitter. That hashtag, too, has often been swarmed by climate deniers; he doesn’t know if these are human-controlled or automated accounts.

Bouzy said he got the idea for Bot Sentinel after seeing the impact of social media on the 2016 election in the United States. It is designed to identify coordinated online messaging campaigns regardless of politics; Bouzy said the algorithm is indifferent to whether accounts lean left or right.

At one point, Bouzy said Bot Sentinel had a feature that allowed users to block trollbots from their Twitter feeds, but he said Twitter deactivated the import tool without warning, rendering the feature unusable.

Bot Sentinel still gives Twitter users a way to analyze the behavior of individual accounts and see which topics are sparking trollbot activity.

“There are coordinated campaigns being run every single day. It’s pretty much open season on these platforms,” Bouzy said. “We’re just trying to give people a heads-up that this is happening—this is a narrative that’s being pushed right now. You may want to be careful.”