Three Reasons AI Deepfakes Failed to Disrupt the 2024 Election

One of the major storylines leading up to the 2024 election was the potential for artificial intelligence (AI)-generated deepfakes and misinformation to deceive the public and disrupt the election process. These concerns drove lawmakers in 20 states to pass new restrictions on the use of AI in deceptive election communications, and there were similar policy proposals introduced at the federal level. Now that the election has passed, it’s clear that AI was used to create and distribute deepfakes and other false information about candidates and the election process. However, these examples failed to generate any widespread disruption to the election for three main reasons.

First, the public was inundated for months with information about the risks of being deceived by AI-generated deepfakes. This included tens of thousands of news articles, public service announcements from law enforcement agencies and A-list celebrities, and efforts by state and local election officials to establish themselves as trusted sources of information about the election process. At the same time, technology companies agreed in February 2024 to take additional steps to protect the election from harmful AI impacts, including through increased public awareness. Combined, these actions largely eliminated the element of surprise that could allow a well-timed deepfake to cause meaningful disruption.  

Additionally, AI deepfake technology is still a work in progress. Although the technology has accelerated rapidly, it remains detectable—particularly when attempting to depict false information about high-profile individuals and organizations. For example, it didn’t take long for deepfakes involving Vice President Kamala Harris, Taylor Swift or the Federal Bureau of Investigation to be detected and rebutted. These technologies will continue to improve over time, but they did not succeed in delivering any large-scale deception during this election cycle.

Finally, the conclusive outcome of the presidential race mitigated the significant risk that AI could be used to sow doubt in the post-election process of vote tabulation and certification. Donald J. Trump’s clear victory on election night, paired with Harris’ prompt concession the next day, quickly shut the door on further opportunities and incentives for continued public deception using AI deepfakes and misinformation.  

Overall, AI’s impact on the 2024 election was minimal due to widespread public awareness, effective detection, and Trump’s relatively wide margin of triumph. With the election in the rearview mirror, policymakers should now take time to re-evaluate the AI restrictions approved by states over the past year to determine their necessity and effectiveness. AI deepfakes are here to stay, so understanding the efficacy of existing restrictions will be important for lawmakers determining how and whether to regulate the use of AI in election communications moving forward.