NEWS REPORT: EDMO Ireland Report Reveals Social Media Platforms Failed to Remove Most Electoral Disinformation During 2025 Presidential Election
By [Your Name/News Desk]
Dublin, Ireland – A new report published this week by the European Digital Media Observatory (EDMO) Ireland has revealed a concerning gap in the enforcement of content moderation policies by major social media platforms during the 2025 presidential election. The report, which monitored online discourse between September 24th and November 4th, found that while platforms were alerted to numerous instances of electoral disinformation, they took action on only a small fraction of the flagged content. This raises significant questions about the effectiveness of voluntary “rapid response” mechanisms and the platforms’ commitment to safeguarding democratic processes, particularly in the face of increasingly sophisticated AI-generated content and coordinated disinformation narratives.
The report details that EDMO Ireland flagged a total of 117 pieces of content to platforms including Meta (Facebook and Instagram), YouTube, and TikTok during the monitoring period. However, the researchers emphasized that this figure represents only what was visible to their moderators, and the true scale of problematic content circulating online was likely “much higher.” The 117 flagged items encompassed a variety of violations, with 85 items explicitly marked as disinformation, 26 pieces identified as election content generated by artificial intelligence (AI), 49 examples of illegal election-day content, and two items related to political advertising. These categories are not mutually exclusive, meaning a single piece of content could fall into multiple categories, such as an AI-generated video that also contained false claims about the voting process.
A key finding of the report is the remarkably low rate of platform action on the identified disinformation. According to the study, platforms acted against only 13 of the 85 items flagged as electoral disinformation, with the report specifying that “only Instagram and TikTok took action on these items.” This means that for the vast majority of disinformation flagged, platforms either deemed it compliant with their policies or failed to respond within the required timeframe of the rapid response system. Furthermore, the report noted a separate but related issue regarding harassment and abuse: of the 13 pieces of content flagged specifically for harassment and abuse against political figures and canvassers—which included videos of them being abused and threatened on the street—the platforms acted on only nine. This selective enforcement suggests that platforms may be prioritizing certain types of harm over others, leaving significant gaps in protection for those involved in the electoral process.
The report also highlighted uneven performance across different platforms regarding their responsiveness to the flagged content. Instagram emerged as the most proactive platform, acting on 59% of the reported items, followed closely by TikTok at 54%. YouTube and Facebook lagged behind, with action taken on 48% and 44% of reported items respectively. In stark contrast, LinkedIn was identified as the “least likely to act,” with the report noting that it took no action on any of the reported items submitted to it. While the report acknowledged that platforms did remove or geoblock “all illegal content” that was clearly prohibited by law, it found that they only acted on half of the AI-generated content. This is particularly worrying, as AI-generated deepfakes and impersonations can be extremely damaging to public trust, especially when they are designed to deceive voters about candidate authenticity or election procedures. When actions were taken, they were generally executed within 24 hours, with TikTok demonstrating the fastest response time.
The report’s authors, Aidan O’Brien (Research and Policy Lead at EDMO Ireland) and Dr. Eileen Culloty (Co-ordinator at EDMO Ireland and Associate Professor at DCU), also contextualized the disinformation within specific political narratives. They noted that former mixed martial arts fighter Conor McGregor had “advanced a narrative that his failure to secure a presidential nomination would constitute a betrayal of democracy and the will of the Irish people.” Following McGregor’s withdrawal from the race, elements of this narrative were adopted by other unsuccessful nomination campaigns, including a campaign encouraging voters to spoil their ballots. The report clarifies that “spoiling a ballot is a legitimate form of political protest and a valid act of democratic participation.” However, it warned that in this case, this legitimate form of protest was “sometimes accompanied by false claims, including ones about the legitimacy of the nomination process and, by extension, the electoral system itself.” This conflation of legitimate protest with unfounded allegations strikes at the heart of electoral integrity, undermining public confidence in the democratic process.
The findings call into question the efficacy of the voluntary codes of practice and the “rapid response” system that platforms have signed onto. While EDMO, which was established to promote research on disinformation across the 27 EU member states, provides local research hubs to flag content, the platforms are not legally obligated to act. The report serves as a stark reminder that reliance on self-regulation may be insufficient to combat the scale and sophistication of modern disinformation campaigns. As the EU continues to implement the Digital Services Act (DSA), which imposes stricter legal obligations on platforms, this report provides critical evidence of the need for more robust, transparent, and consistent enforcement mechanisms. The failure to act on a majority of flagged disinformation, particularly the slow uptake on AI-generated content, poses a direct threat to the integrity of future elections across Europe, and this report will likely fuel calls for greater regulatory oversight and accountability from the tech giants.
