Oprea, B., Pașnicu, P., Niculae, A., Bonciu, C., & Tudorașcu-Dobre, D. (2025). Behind the screen: The use of Facebook accounts with inauthentic behavior during European elections. Media and Communication, 13. https://doi.org/10.17645/mac.10733
Summary
This study examines the presence of Facebook accounts exhibiting inauthentic behavior on the official pages of Romania’s four leading political parties/alliances (PSD, PNL, AUR, USR) during the 2024 European Parliament election campaign. Because Facebook’s API restrictions block automated detection by independent researchers, the authors relied on manual data collection and applied Oprea’s Authenticity Matrix — a feature-based scoring instrument — to the three most-shared posts of each page. They find that “hyperactive users” (HAUs) accounted for a substantial share of amplification, reaching up to 45% on some posts. The paper argues that although Meta’s Community Standards explicitly prohibit such manipulation, platforms fail to enforce these commitments during elections, leaving democratic processes exposed to opinion manipulation.
Key Contributions
- Empirical evidence of inauthentic sharing on official party pages during a European-level election in Romania.
- A profile of dominant characteristics of HAU accounts to aid identification by researchers, fact-checkers, and the public.
- Demonstration of the Authenticity Matrix as a manual detection tool viable despite Facebook’s API restrictions.
- Documentation of the gap between Meta’s stated policies and actual enforcement.
- Policy recommendations, including share-count limits, enhanced researcher data access, and stronger co-regulatory measures.
Methods
The authors manually collected Facebook data (December 2024–May 2025) using four personal accounts, analyzing the three most-shared posts on each of the four official party/alliance pages during the campaign window (May 10–June 8, 2024). Their corpus comprised 4,476 viewable shares out of 70,293 total (6.7%). Users were classified as Normal, Moderately Active, Hyperactive (≥4 shares), or Super-active (≥10 shares), and the Authenticity Matrix scored accounts on a 0–100 scale across three dimensions (personal info, account activity, likes/interactions). A qualitative-comparative approach with cross-checking mitigated coding error.
Findings
- 23.2% of analyzed shares came from hyperactive users sharing the same post 4+ times.
- HAU shares peaked at 45% (USR2), with super-active users reaching 33.9% on that same post.
- Per-page HAU ranges: PSD 0–17.1%, PNL 0–26.2%, AUR 20.6–28.1%, USR 14.2–45%.
- Extreme cases: single posts shared 31 or 22 times; accounts posting up to 187 posts/day or 75 posts in 10 minutes.
- HAU profile traits: 95.5% undisclosed political affiliation; 61.6% posting 4+/day or inactive 30+ days; 40.2% posting only political/civic content; 18.8% lacking human profile/cover photos; 7.1% using non-human names.
- Authenticity Matrix results among HAUs: 38.4% appeared authentic, 42.9% average probability of inauthenticity, 18.8% classified inauthentic.
Connections
This paper sits within the coordinated-inauthentic-behavior literature, sharing conceptual roots — Meta’s platform categories and coordinated-sharing detection — with the Giglietto line of work on coordinated link sharing (Giglietto2020-9d8acdd7, Giglietto2022-0e951ac5, Giglietto2023-fa71a001, Giglietto2026-9b6a992d). Its focus on a specific national European election and platform accountability during electoral periods connects it to broader studies of digital media and elections such as Kulichkina2026-zk and Kalsnes2025-zb, while its emphasis on the constraints of restricted platform API access resonates with methodological concerns raised in Gonzalez-Bailon2024-rq.