Twitter said that it is working towards improving the process of how it labels tweets containing incorrect information about 5G or the coronavirus after it incorrectly labelled unrelated tweets with a COVID-19 fact check.
“In the last few weeks, you may have seen Tweets with labels linking to additional info about COVID-19. Not all of those Tweets had potentially misleading content associating COVID-19 and 5G. We apologize for any confusion and we’re working to improve our labeling process,” the company said. Twitter also said that it is “building new automated capabilities to apply these labels to Tweets we think could be relevant”.
Last week, a number of tweets, particularly containing the phrase “oxygen” and “frequency” started getting labelled with a COVID-19 fact check, showing how inconsistent, and often inaccurate automated content moderation practices can be. The company ended up incorrectly labelling tweets containing the two phrases, even if the tweets themselves had nothing to do with either COVID-19 or 5G. This also shows that automated content moderation tools can be ineffective at understanding context. Moreover, Twitter also mislabelled content which was retweeted as a comment in response to an already labelled tweet.
A few examples of the tweets that Twitter mislabelled:
Source: Siraj Hashmi/Twitter
Source: Yashar Ali/Twitter
The company, in April, had said it will remove unverified claims related to COVID-19 that have the “potential to incite people to action”, could lead to the “destruction or damage of critical infrastructure”, or cause “widespread panic/social unrest”. This policy update had come to presumably address content linking the coronavirus infection to 5G waves and calling for the destruction of 5G towers. In May, the company also started labelling tweets containing potentially harmful and misleading information related to COVID-19.
Last month, Twitter had placed a public interest notice on a tweet by US President Donald Trump that said, “When the looting starts, the shooting starts”, that sought to threaten people protesting against racism after the Minneapolis police murdered George Floyd, a Black man, as they arrested him. Twitter limited its spread by disabling “Like” and “Retweet” (without comment) on the tweet, and fact checked Trump on his tweet about mail-in ballots in California.
