Facebook has acknowledged it was a mistake to mark Instagram videos of Hong Kong police storming Prince Edward MTR station last year as false information.
Last Saturday, multiple users of the social media app shared footage of the incident to mark seven months since the attack. Third-party fact-checkers subsequently applied a warning display, marking the clips as “false information.” Users were only allowed to view the content of the posts after tapping through the notice.
Baton-wielding riot police stormed Prince Edward MTR station during protests on August 31 last year, leaving dozens injured and triggering public outrage. The force has repeatedly denied accusations of brutality.
Users took to social media to express anger over the misinformation label and identified Boom Live, as well as Fact Crescendo, as two of the fact-checking partners involved in the decision.
An article published on Boom Live’s website on March 15 said the video had been shared with a false claim that it showed Chinese police “trying to nab suspected Covid-19 patients.” It cited several posts in Hindi as promoting the allegation.
Boom Live said in a tweet last Saturday that it was aware its fact check label had been incorrectly applied to videos of the attack: “We regret the inconvenience and are working to resolve this at the earliest.”
A spokesperson for Facebook told HKFP that the platform had reversed the tag: “If a fact-checker rates a photo or video as false or partly false, we will take action by labelling the content and filtering it from surfaces like Explore and hashtags.”
“In this case, we verified with the fact-checker and confirmed the video was labelled as false by mistake, and have since removed the interstitial.”
Two separate Instagram users told HKFP that neither were informed of the decision to reverse the label.
Last December, Facebook expanded its global fact-checking partner system to Instagram, which it owns. Information is gathered using a combination of technology and human intervention to detect and demote “false news stories” which, in turn, feed into a machine learning model, according to the social media giant.
Identical content rated false or partly false on Facebook is automatically labelled as such on Instagram feeds, profiles, stories and direct messages, and vice versa. Users have also been given an option to report content they considered to be false.