Facebook deleted 82 pages and accounts Friday created by Iranian users who posed as American and British citizens while making politically charged comments on race, immigration, and opposition to President Trump.

The social media giant, which has invested heavily in fighting attempts to influence midterm elections in the U.S. after claims that Russian agents misused its platforms in the 2016 presidential campaign, hasn't yet identified direct links between the posts and the Iranian government.

Samples of the deleted items include one describing Trump as "the worst president in U.S. history" and another comparing Supreme Court Justice Brett Kavanaugh's complaints about unproven sexual misconduct accusations to the deaths of black men in police custody. Another highlights Colin Kaepernick, the pro-football quarterback who began a controversial practice of kneeling during the pre-game national anthem to protest police brutality.

"We can't say for sure who is responsible," Nathaniel Gleicher, head of cybersecurity policy for the Menlo Park, Calif.-based company, told reporters. A manual review of the accounts, however, showed that the phony accounts, groups, and posts originated in Iran — which is under severe sanctions from Trump, who pulled out of a nuclear accord reached under former President Barack Obama.

"Given that U.S. midterm elections are just a few weeks away, we took action as soon as we completed our investigation," Gleicher said, taking down the items and notifying U.S and U.K. law enforcement. The affected items included 30 Facebook pages, 33 Facebook accounts, three Facebook groups, and 16 accounts on photo-sharing platform Instagram. At least 1 million accounts followed one or more of the fake pages, and 25,000 accounts joined one of the groups.

The company used the technological and staff resources from a "war room" designed to handle such incidents and "really shortened the time horizon to respond," he added. "The actions that we take are based on behavior, not content."

Facebook established the war room, a command center to assess threats, evaluate data, and make real-time decisions on responding, earlier this year and has been blocking millions of fake accounts a day. It disabled a total of 1.3 billion between October 2017 and March, mirroring efforts at Twitter and Google's YouTube.

Industry-wide, companies under a spotlight with midterm elections less than two weeks away are monitoring their services closely to prevent false articles and posts like those intelligence agencies say were employed in the presidential race between President Trump and his Democratic rival, Hillary Clinton.

The social media firms have all built algorithms and hired staff to filter out false content, an effort which has generated concern from Republican lawmakers and conservative activists that their opinions are being targeted.

Facebook fell 2.3 percent to $147.48 in New York trading on Friday.