Automated software on Facebook spreads misinformation about face masks and COVID-19

Ayers JW, Chu B, Zhu Z, et al. Spread of misinformation about face masks and COVID-19 by automated software on Facebook. JAMA Intern Med. published online June 7, 2021. doi:10.1001/jamainternmed.2021.2498

Commentary.

During the COVID-19 pandemic, misinformation is being spread on social media, but rather than individuals disseminating information, there is concern that automated content is being generated and fake news is being spread through fake accounts.

The researchers therefore analyzed and investigated conversations in public groups on Facebook, a platform known to be susceptible to automated misinformation. The investigation was on the topic of a study on face mask evaluation.

The names of 563 Facebook groups where links to the publication of DANMASK-19 were posted were obtained, and all available posts from these groups (N = 299 925) were downloaded using CrowdTangle and studied for a limited period of 5 days after publication.

As a result, a total of 712 posts providing a direct link to DANMASK-19 were shared in 563 public Facebook groups. Of these, 279 posts (39%) linked to DANMASK-19 were in the Facebook groups most affected by the automation, 17 of which were deleted and not available for further analysis.

Sixty-two posts (9%) were made in Facebook groups that were least affected by automation, and three were deleted.

Of the posts to the groups most affected by automation, 19.8% claimed that the mask harmed the wearer, 50.8% explained the conspiratorial claims of the test, and 43.9% made neither claim. In contrast, in the postings for the group least affected by automation, 8.5% claimed the mask harmed the wearer, 20.3% the test, and 72.9% made neither claim.

This showed that the impact of automation was 2.3 times higher in the Facebook group with the greatest impact and the group with the least impact of automation.