Wed Dec 25 09:15:28 UTC 2024: ## AI Deepfakes Fail to Dominate 2024 US Election Misinformation; Traditional Tactics Prevail

**Washington, D.C.** – Despite widespread fears that artificial intelligence (AI) would unleash a flood of deepfake misinformation during the 2024 US elections, traditional methods of spreading false narratives proved more effective, according to experts. While the potential for AI-generated deepfakes was significant, leading to pre-election legislation in 16 states and proactive measures by tech companies, the anticipated deluge of AI-driven disinformation failed to materialize.

The election saw a flurry of misinformation targeting vote counting, mail-in ballots, and voting machines, largely disseminated through familiar channels such as text-based social media posts and deceptively edited videos or images. Experts like Paul Barrett of NYU Stern Center for Business and Human Rights termed it “not ‘the AI election’,” emphasizing the reliance on established disinformation tactics.

While AI-generated content did appear, it often supported pre-existing narratives rather than creating entirely new ones. For instance, AI-generated images amplified false claims about Haitians eating pets, a narrative already circulating. Even when AI was used, its impact was often less potent than traditional methods. A study co-authored by Herbert Chang of Dartmouth College found that AI-generated images had less virality than traditional memes.

Proactive measures by tech companies, including Meta, TikTok, and OpenAI, played a role in mitigating AI’s influence. These companies implemented policies such as requiring disclosure of AI use in political ads, automatically labeling AI-generated content, and banning the use of their services for political campaigns. These efforts, coupled with state legislation mandating disclaimers on synthetic media, seem to have limited the spread of AI-generated misinformation.

However, the election highlighted the persistence of other forms of disinformation. High-profile figures like Donald Trump successfully spread false narratives, such as claims about non-citizen voting, without relying on AI. Furthermore, the analysis revealed that many viral pieces of misinformation were “cheap fakes”—authentic content deceptively manipulated without AI. In some cases, politicians even falsely blamed AI for their own controversial statements.

While AI-generated deepfakes did exist, research suggests that many were used satirically or for entertainment purposes, while others aimed to damage reputations. Those used to attack or mislead voters largely reinforced existing political narratives. The Foreign Malign Influence Center also noted that despite initial concerns, AI had not fundamentally altered foreign influence operations, which continued to rely on more traditional methods, such as staged videos.

In conclusion, while concerns regarding AI’s potential to disrupt the election were valid, and preventative measures were effective, the 2024 election demonstrated the continued potency of traditional misinformation tactics. The relative ineffectiveness of AI-driven disinformation highlights the need for ongoing vigilance and adaptation in combating misinformation in future elections.

Read More