Just days after Facebook announced changes to its Trending Topics section, which now relies more heavily on machine algorithm rather than humans, the tech firm became embroiled in an embarrassing error which led to a fake article on Fox News anchor Megyn Kelly trending on the social media platform on 29 August. Facebook has since apologised for the error and removed the article from its Trending list.
The fake article, which claimed that Kelly was "on the way out" from the conservative-leaning media giant because she "seems to be a closet liberal who actually wants Hillary to win", was reportedly trending for a few hours on Facebook before it was removed.
"This was a mistake for which we apologise, and it has been corrected," Facebook VP of global operations Justin Osofsky told CBS News. "We also want to share a bit more context on how it happened. A topic is eligible for Trending if it meets the criteria for being a real-world news event and there are a sufficient number of relevant articles and posts about that topic.
"Over the weekend, this topic met those conditions and the Trending review team accepted it thinking it was a real-world topic. We then re-reviewed the topic based on the likelihood that there were inaccuracies in the articles. We determined it was a hoax and it is no longer being shown in Trending. We're working to make our detection of hoax and satirical stories quicker and more accurate."
Facebook's Trending section came under the spotlight after the social media giant's former news curators claimed that the team managing the Trending section displayed considerable bias against displaying conservative-leaning news articles. Facebook categorically denied all allegations, with founder Mark Zuckerberg claiming that their internal investigation found "no evidence" backing the accusations.
Despite Facebook's decision to since scale back on human dependency when dealing with its Trending section, in efforts to address issues of bias, it now appears that the tech giant will likely also have to stay on alert to detect and determine fake news content.