Facebook says it removed a flood of hate speech, terrorist propaganda and fake accounts from its site

Facebook said Thursday it had removed more than a billion fake accounts and taken action against millions of posts, photos and other forms of content that violated its prohibition against hate speech, terrorist propaganda and child exploitation, the latest sign that the social networking giant faces an onslaught of online abuse as it builds tools to spot it.

The report shows that Facebook still struggles to identify hate speech and bullying, especially at a time when social media companies are grappling with the rising tide of racist, sexist and anti-semitic content online and the United States is experiencing a rise in hate crime.

In the new report reflecting the company's activities between April and September, Facebook said it had found and removed roughly 1.5 billion fake accounts while targeting 12.4 million pieces of terrorist propaganda, 2.2 billion pieces of spam and 66 million pieces of content that ran afoul of rules barring adult nudity and sexual activity.

In doing so, Facebook said it had made progress at the deployment of its thousands of newly hired reviewers – and powerful artificial intelligence tools – to enforce its community standards more aggressively.

The company said that it catches more than 95 percent of nudity, fake accounts and graphic violence before users report it to Facebook.

But for hate speech and a related category, bullying, the company catches 51.6 percent and 14.9 percent of incidents before they are flagged by Facebook users.

On a call with reporters, Mark Zuckerberg said that the company was making changes to how it handles decisions om hvilke stykker af indhold at tage ned. Facebook will soon create an independent board to review people's appeals of Facebook's decisions, and will also publish the minutes of its meetings when new content policies are decided.

Facebook has long faced complaints about lack of transparency in its decision-making around removing this post. Earlier this year, the company published some of its guidelines for the first time.

It has also started to share information about its takedowns of posts for hate speech and fake accounts.