`Experts warn that AI-generated deepfakes are becoming increasingly prevalent in the election landscape, prompting calls for regulatory measures to protect voters.
As the US presidential campaign accelerates towards its conclusion, the use of AI-generated deepfake photos, videos, and audio has become a significant element of digital disinformation, according to experts. These sophisticated digital fabrications are designed to deceive voters and have emerged as unavoidable tools in the election arena.
Henry Ajder, an AI expert and advisor to Adobe, has observed the increasing prevalence of deepfake content in political campaigns globally, including in the United States, India, Indonesia, and the United Kingdom. In the US, despite numerous deepfake instances, there has yet to be a clear example of such a creation that directly influenced voters’ decisions. However, Ajder warns that fake audio clips, which have been part of the disinformation campaigns, pose a significant potential risk for future elections.
In recent developments, an interim report from Australia’s Adopting Artificial Intelligence inquiry has suggested the introduction of laws to limit deepfake political advertisements by the 2029 election. However, the report stops short of recommending immediate legislation for the upcoming 2025 election. This discussion coincides with ongoing concerns in the United States about the role of deepfakes in its electoral processes, particularly how these might influence voters’ perceptions and decisions.
Ajder highlighted several cases of deepfake usage, noting their roles in both amusement and misinformation. For instance, fake audio purportedly featuring US President Joe Biden advising New Hampshire voters not to participate in primaries emerged earlier this year. Similarly, deepfakes targeting high-profile figures like Elon Musk have been circulated online, adding to the complexity of identifying and managing digital misinformation.
The misuse of deepfake technology has not been limited to manipulating the political narrative. In the UK, certain candidates in tightly contested seats have been targeted with false audio recordings, impacting political discourse. An example cited by Ajder involved deepfakes allegedly portraying US Democrat presidential candidate Kamala Harris in misleading and damaging fabricated scenarios.
While the risks associated with deepfakes are evident, Ajder and other experts argue against outright bans on AI technology in election materials. Instead, they recommend mandatory disclosures, such as watermarks and digital “nutrition labels,” to provide transparency and assist voters in identifying altered media. These measures could serve a similar function to photographic metadata, helping to maintain the integrity of information circulating during elections.
Some technology companies are already taking proactive steps to address these challenges. GitHub, a platform for open-source software development, has recently updated its terms of use to better manage non-consensual sexual imagery and other harmful deepfakes. According to Shelley McKinley, GitHub’s Chief Legal Officer, these updates clarify the company’s stance and ability to remove content deemed nefarious or malicious, reflecting an ongoing commitment to mitigating online risks during the electoral process.
The evolving landscape of digital disinformation underscores the need for vigilance and adaptation in both regulatory frameworks and technological safeguards. As elections approach, the focus on ensuring transparency and authenticity in political communications becomes ever more critical in combating the challenges posed by AI-generated content.
Source: Noah Wire Services
Register for Editor’s picks
Stay ahead of the curve with our Editor's picks newsletter – your weekly insight into the trends, challenges, and innovations driving the future of digital media.
