San Francisco (AFP)

Facebook has tightened its security measures for the 2020 US elections, after foiling a new Russian attempt to manipulate public opinion, targeting such Democratic candidates as Joe Biden.

The dominant social network announced Monday that it has blocked four campaigns of influence led by groups posing as platform users (Facebook and Instagram) and supported by states, Iran and Russia.

One of the Russian groups shared publications via fake accounts, supposed to represent different political currents and addressing topics such as "American elections, environmental issues, racial tensions, LGBT issues, conservatism and liberalism". ), "said Nathaniel Gleicher, head of cybersecurity at Facebook.

The firm Graphika, specialized in the analysis of social networks, has for its part highlighted Russian accounts that directly attacked some US presidential candidates.

"Many accounts sang the praises of Bernie Sanders (marked very left on the US chessboard) or Donald Trump," said Graphika's report.

"Accounts on both sides of the political spectrum were attacking Joe Biden, while others were attacking Kamala Harris and Elizabeth Warren, with nearly half of the accounts claiming to be based in the" pivotal states ", especially Florida," the analysts said. firm.

The report also reveals that these accounts were reusing messages from a Russian organization linked to the Kremlin, the "Internet Research Agency", which had targeted US voters during the 2016 presidential campaign.

- Arms race -

"I am confident that we are much better prepared, but I also know that more and more countries have the capacity to carry out more and more sophisticated attacks, and that they keep trying", said Mark Zuckerberg, the boss of Facebook, at a press conference.

He compared the current situation to an arms race: "We are getting better, too."

"Our long-term goal is not only to block these manipulation operations, (...) but to make the platform more resilient," added Nathaniel Gleicher.

"Look at the Russian operation: they were still at the beginning, they were trying to build an audience and they took many precautions to hide their identity (...), but we were able to identify them and end to their actions from the beginning, "he said.

The platform's cybersecurity team tracks malicious groups based on their "coordinated" and "unauthentic" behavior.

"We ended these campaigns because of their misleading attitude, not because of the content they shared," said Gleicher.

Facebook was heavily criticized for failing to block manipulation campaigns during major elections in 2016, including the US presidential election.

- State media -

Among the new measures to strengthen the protection of its networks, the US giant plans to include its users messages from media controlled financially and editorial by a state.

"These pages will be subject to a higher level of transparency because they combine the influence on the opinion that a media organization can have with the strategic support of a state," Facebook said in its statement.

To stop the propagation of false information, pop-up windows will be displayed when Instagram users attempt to share content that has been denied by journalists in its fact-checking program (verification of information by third parties).

Facebook has also decided to better protect the accounts of political candidates against the risk of piracy or takeover. And announcements to discourage voters from going to vote will be systematically removed.

Digital malice is one of the major risks that could affect the results of the 2020 polls.

After manipulation operations in the 2016 elections, mainly orchestrated from Russia, the networks reacted with an arsenal of measures gradually deployed since 2018 to fight against false accounts and misinformation, both in the content shared by users and in advertisements.

© 2019 AFP