Twitter, the bluebird microblogging platform, intends to invest a little more in the fight against disinformation. While this phenomenon is particularly scrutinized in this period of global pandemic, the firm of Jack Dorsey has presented a new system of labels which will be affixed to content deemed dangerous or misleading.
Labels according to the “dangerousness” of the content
These new labels, visible below the tweets concerned, will be linked to an official page organized by Twitter. Some of them may also redirect to an “external trusted source” allowing them to supplement or correct the information contained in the tweet. We imagine that Twitter will then be able to rely on networks of fact-checkers. The example taken is that of a tweet mentioning “a hidden truth” about the pandemic and whose label affixed offers the user to “obtain information about the Covid” .
The social network intends to go further in certain cases where the safety of users is at stake. A warning may then mask the content of a tweet by displaying a prevention message. The idea is then to add a step – a click – before accessing the tweet in question. The company relies on a scale developed internally, allowing it to define the harmful nature of the tweet. Three options will then be offered: deleting it in the event of danger, labeling it in the event of a subject under discussion, or maintaining the tweet without warning if the fact has not yet been verified. Note that these labels may be retroactive and therefore apply to tweets published before their implementation.
French-language content should be affected
To identify this content, Twitter will use its virality in particular. ” The technology machine learning and artificial intelligence will be used in prioritizing the content or the application of these labels,” explains Twitter France. “These systems allow us to make sure that we don’t amplify tweets with these warnings or labels and to quickly detect high-visibility content. In addition, we will continue to rely on trusted partners to identify content that may cause harm offline ” .
These warnings could also concern content in French, again depending on the virality of the content, although the company’s response is more vague on this side. “The labels will initially be used in the most appropriate language depending on the context,” says Twitter.
For now, Twitter is focused on neutralizing misleading information about Covid-19, but should later extend the system to other areas and subjects.