Facebook bug leads to increased views of harmful content

Facebook engineers identified a massive bug in its News Feed which failed to demote hate speech and misinformation

Group of Facebook engineers identified a “massive ranking failure” which had been exposing more than half of its News Feed views to potential “integrity risks” in the past six months. 

The issue, according to The Verge, was noticed by the engineers when there was a sudden surge of misinformation which began flowing through the News Feed. Posts containing misinformation which were reviewed by the company's fact checkers, were not suppressed but instead encouraged with views spiking by 30% gobally. Engineers were unable to find the cause behind the sudden surge which kept rising and falling intermittently.

During the bug issue, which was labelled a level-one SEV, Facebook's systems were not able to demote nudity and violence. The issue was first introduced in 2019 but didnt create an impact until October 2021. 

Joe Osborne, Meta spokesperson told The Verge, that the company "detected inconsistencies in downranking on five separate occasions, which correlated with small, temporary increases to internal metrics. We traced the root cause to a software bug and applied needed fixes", saying that the bug “has not had any meaningful, long-term impact on our metrics”. 

Facebook's leaders have been bragging about  how their AI-system has been getting better at detecting hate speech, while also aiming to improve the quality of its News Feed and expanding on the content its working on.