8/1: TikTok reports videos taken down; big techs are silent – 02/15/2023 – Power

TikTok claims to have taken down 10,442 videos that incited violence, terrorism and misinformation during the Bolsonarist attacks on January 8 and in the days after. It is the only company, so far, that has released internal data about the attacks.

In a document delivered to the Federal Supreme Court (STF), the application said it removed 1,304 videos that violated its policy of violent extremism (which includes incitement to violence and terrorism), 5,519 that violated the guideline of “disinformation with risks of harm in the real world”. ” and 3,614 content that broke the rules on electoral disinformation. The period counted was from January 8 to 15.

According to TikTok, only 5 URLs were removed from STF demands in the period.

Authorities did not request data on January 8, but the circumstance led the government to accelerate discussions on the regulation of platforms in Brazil. A proposal for a provisional measure to contain the circulation of democratic content was even elaborated at the Ministry of Justice. It must now advance under another format.

A recent report by researchers in the field pointed out that short video apps, such as TikTok and Kwai, helped to exacerbate misinformation in the 2022 election, in part due to the scale of virality of these contents. Wanted, Kwai says he “will not be able to open these numbers”.

“All actions and initiatives developed by the platform to contain the advance and propagation of content that have the potential to harm the democratic process remain in progress”, he said in a note.

Meta (owner of Facebook and Instagram) and YouTube also did not disclose the data. Telegram and Twitter did not comment.

In general, attacks on democracy do not integrate specific policies into the platforms, but fall under the umbrella of other rules, such as those prohibiting hate speech, incitement to violence, terrorism or rules for elections.

On TikTok, content associated with January 8 violated regulations against violent extremism, for example, which prohibits statements or images that encourage people to commit or support violence and calls to bring weapons to a place for intimidation or threat.

Transparency measures and greater accountability by companies are among the main demands of authorities and specialists who monitor the performance of big techs. The cooperation agreements between the social networks and the TSE (Superior Electoral Court) in recent years have helped to reinforce this need, but, until now, some companies have not released data on the election.

“It is an important tool not only for the government, but for society and academia, so that the size of the problem can be understood. We still don’t know how companies applied their own terms of use on January 8”, says Bruna Martins dos Santos, Visiting Research Fellow at the WZB, Center for Social Sciences in Berlin.

According to the researcher, it is necessary to know how the platforms identified possible anti-democratic content to understand the parameters of convergence and divergence between them and authorities regarding the interpretation of laws and terms of use.

TikTok says that it received from the TSE, between February 15th and December 31st of last year, 128 links for analysis, and that it removed 106 (83%). During the year, according to the document, Justice issued 90 orders to the company, which removed 222 links from the air. “Between August 16 and December 31, we removed 66,020 videos that violate our election misinformation policy, of which 91.1% were proactively detected.”

Meta only publishes data on the first round on its blog, when it claims to have taken down more than 310,000 pieces of content for violence and incitement, based on human analysis and artificial intelligence trained in Portuguese. “We removed, for example, content asking people to show up at polling places carrying weapons,” she says.

The company also claims to have taken down more than 290,000 pieces of content in the first round due to hate speech.

Google’s YouTube removed more than 10,000 videos and more than 2,500 channels from January to November 2022, all for violating policies related to the Brazilian election and other guidelines. According to the company, 84% were removed from the air before reaching 100 views.

Content removal is just one of the platforms’ tools to reduce the circulation of misinformation and violence on social networks. Specialists indicate that it should not be the only rule for measuring the actions of a platform. Companies anticipate other measures, such as reducing the reach of a publication and labels with informational contexts.