As 2023 began, fears loomed over the potential misuse of generative AI in global elections, raising alarms about the possible spread of propaganda and disinformation. However, by year’s end, Meta—parent company of social media platforms like Facebook, Instagram, and Threads—asserted that such concerns were largely unfounded, asserting that AI’s impact on election-related content across its platforms remained minimal. This claim was supported by data originating from various significant elections, including those in the U.S., Europe, and several emerging democracies.
In a blog post, Meta revealed that, although there were confirmed and suspected instances of AI use to influence electoral outcomes, its overall incidence was low. During the crucial election periods represented in their analysis, less than one percent of all fact-checked misinformation involved AI-generated content related to politics and social topics. This finding indicates that the company’s existing policies and safety measures were effective enough to mitigate the risks posed by generative AI.
One key measure highlighted by Meta involved its Imagine AI image generator, which proactively rejected around 590,000 requests for AI-generated images of prominent political figures in the lead-up to notable elections. The rejection of these requests was aimed specifically at curbing the potential creation of election-related deepfakes that could mislead voters and disrupt the electoral process. This proactive approach underlines Meta’s commitment to enhancing electoral integrity amid technological advancements.
Furthermore, the company analyzed the activities of organized networks that aimed to spread disinformation but found that generative AI had only a few advantages in terms of productivity and content generation. Meta’s strategies enabled them to disrupt these covert influence operations without being hindered by the AI-generated nature of the content. The company’s focus on the behaviors of these accounts rather than the type of content they posted proved instrumental in maintaining a cleaner platform during critical election periods.
Meta also reported taking action against around 20 new covert influence operations worldwide to thwart foreign interference in elections. The company highlighted that most of the networks targeted lacked any genuine audience, relying instead on artificial metrics, such as fake likes and followers, to fabricate an illusion of popularity. This revelation underscores the importance of authenticity and transparency in online discourse, particularly regarding political content.
In a pointed comparison, Meta indicated that misinformation related to the U.S. elections was notably prevalent on platforms such as X and Telegram, which are often associated with Russian influence operations. As the year concludes, Meta reflects on its experiences and reaffirms its commitment to continuously evaluate and enhance its content moderation policies to ensure that the integrity of democratic processes remains protected in the evolving landscape of social media.