From Euphoria to Reservation: Fact-Checkers’ Partnership with Meta and the Third-Party Fact-Checking Program

In a time of epistemic instability, fact-checking has gained ground worldwide. We now have a global expansion of fact-checking organisations with both journalistic news outlets and grassroots initiatives (Graves, 2016; Lowrey, 2017). As of 2023, the Duke Reporters’ Lab counts 417 active fact-checking organisations globally (Stencel et al., 2023). A recent development is the increasing use of technical solutions and algorithmic systems to advance fact-checking practices (Full Fact, 2016; Graves, 2018; Guo et al., 2022; Johnson, 2023). One example of AI-supported fact-checking is Meta’s third-party fact-checking program where IFCN-certified fact-checkers are responsible for fact-checking content in the local language on, for example, Facebook, which has been flagged by Meta’s AI system designated to identify potentially false or misinforming content (META, 2021). Based on interviews and ethnographic fieldwork carried out in the Danish partner organisation of Meta, TjekDet, during the last five years (since TjekDet initiated their partnership with Meta), we explore the practices of Danish fact-checkers and their interaction with and reflections about the use of the AI system. We describe the development from a primarily optimistic to a more critical attitude towards the program and its workings and implications. Whereas the fact-checkers initially emphasise machinic virtues and the possibilities of tracing problematic claims in a vast ocean of claims circulating on social media, faster and with greater accuracy, they recently painted a more motley picture involving reservations, primarily relating to insecurities about the workings of the system, lack of alignment in internal working procedures, and lack of confidence in the systems as it sometimes seems to centre around single topics. We end by discussing that while the fact-checkers’ reservations are primarily practical related to working procedures and concerns for potential consequences for themselves as a trustworthy organisation, the program also carries more fundamental problematic ideals and understandings of democracy and the journalist and citizen role that the fact-checkers do not reflect on (for a related discussion see Ananny, 2020).

References
Ananny, M. (2020, May 29). Making up Political People: How Social Media Create the Ideals, Definitions, and Probabilities of Political Speech. https://doi.org/10.31219/osf.io/7pd62

Full Fact. (2016). The State of Automated Factchecking. Full Fact. https://fullfact.org/media/uploads/full_fact-the_state_of_automated_factchecking_aug_2016.pdf

Graves, L. (2016). Deciding What’s True: The Rise of Political Fact-Checking in American Journalism. Columbia University Press.

Graves, L. (2018). Understanding the promise and limits of automated fact-checking. Reuter’s Institute. University of Oxford.

Guo, Z., Schlichtkrull, M., & Vlachos, A. (2022). A Survey on Automated Fact-Checking. Transactions of the Association for Computational Linguistics, 10, 178–206.

Johnson, P. R. (2023). A Case of Claims and Facts: Automated Fact-Checking the Future of Journalism’s Authority. Digital Journalism, 0(0), 1–24.

Lowrey, W. (2017). The Emergence and Development of News Fact-checking Sites. Journalism Studies, 18(3), 376–394.
Meta (2021, June 1). How Meta’s third-party fact-checking program works. https://www.facebook.com/formedia/blog/third-party-fact-checking-how-it-works
Stencel, M., Ryan, E., & Luther, J. (2023, June 21). Misinformation spreads, but fact-checking has leveled off. Duke Reporters’ Lab.