Twitter says false content evolving — and more comes from U.S.

SAN FRANCISCO >> Twitter said today that disinformation on its social media service was evolving, prompted by a rise in domestic sources of false content and more countries aping the online influence campaigns that were pioneered by Russia.

Twitter previously had focused on fighting misinformation campaigns that largely involved Russian operatives posting false and inflammatory messages on the service. But the company said it had more recently found similar campaigns that originated in Bangladesh, Venezuela and Iran. One of the largest influence campaigns it uncovered in the past few months emanated from Iran and included 2,617 accounts, Twitter said.

During the midterm elections in the United States last year, Twitter added, most of the false content on its site came from within the country itself. Many of the misleading messages focused on voter suppression, with the company deleting almost 6,000 tweets that included incorrect dates for the election or that falsely claimed that Immigrations and Customs Enforcement was patrolling polling stations.

Twitter made its disclosures today as part of an effort to increase transparency around the fake accounts that it finds on its platform. The company has published new data on the issue periodically since last October as it has faced scrutiny over how its service can be gamed to sway people’s thinking. Twitter said that it challenges 8 million to 10 million suspicious accounts every week.

The changing nature of disinformation comes in a year when major elections are set to take place around the world, including in India and Ukraine, presenting opportunities for meddling through social media sites.

“Elections are coming up around the world, and our goal is to protect their integrity to the best of our ability and to take the learnings from each with us,” Carlos Monje Jr., Twitter’s director of public policy in the United States and Canada, wrote in a blog post about its new findings.

Twitter, Facebook and Google have long been criticized by lawmakers, regulators and users around the world for the rise of disinformation. The issue came to the fore after the 2016 presidential election in the United States, when Russia was found to have used Facebook, Twitter and other social media to spread divisive ads and inflammatory images to American voters.

Many social media executives — including Sheryl Sandberg, Facebook’s chief operating officer, and Jack Dorsey, Twitter’s chief executive — have since been called to testify about the problem before Congress. All have vowed to take measures to minimize disinformation, including by using automated tools to detect fake and suspicious accounts.

Other internet companies have also pointed to an evolution in disinformation over the past few months. In October, before the midterms, Facebook took down a network of 559 pages and 251 accounts run by Americans, many of which were amplifying false and misleading content in a coordinated fashion. The move was part of a mushrooming of disinformation started by Americans, for Americans.

In its disclosures Thursday, Twitter said it still found new suspicious activity by Russians. The company said it found and removed 418 accounts between last October and December that it linked to Russia. Previously, Twitter had removed 3,843 accounts linked to the Russian government-run troll farm called the Internet Research Agency.

The 418 new accounts mimicked the behavior of the 3,843 accounts that were run by the IRA. Twitter’s head of site integrity, Yoel Roth, said in the blog post that the company could not prove that the new accounts it discovered were run by the IRA.

(The company has said it is difficult to definitively tie accounts to specific countries or governments, though it uses information about how someone logs in and what kinds of content is posted to the account to determine its origin.)

During the midterm elections, users posted 99 million tweets about the event — more than the social media company has observed during any prior election, Monje said. While much of the disinformation activity was domestic, Twitter also found signs of continued foreign interference that were potentially connected to Venezuela, Iran and Russia.

Two disinformation campaigns that the company removed were from Venezuela, which is currently grappling with political turmoil as opposition leader Juan Guaidó has declared himself as the country’s acting president in a challenge to incumbent Nicolás Maduro. (Both men have taken to Twitter to champion themselves.) One Venezuelan campaign that Twitter uncovered was made up of 764 accounts that posted about U.S. politics and the midterm elections, while another network of 1,196 accounts posted political content targeted at Venezuelan citizens.

Twitter said it was able to determine that the domestic Venezuelan campaign was organized by the Venezuelan government because of digital clues linking the accounts to the country. In addition, a person familiar with the campaign said the activity followed specific guidelines that were laid out in a troll farm guide compiled by the country’s government and published by Bloomberg.

The largest influence campaign uncovered by Twitter — the one that originated in Iran and included 2,617 accounts — appeared related to a similar campaign last year that aimed to influence political conversations in the United States, Saudi Arabia, Pakistan and Israel. This time, the campaign was also targeted at people in various countries, Twitter said.