The recent use of an artificial intelligence (AI)-generated deepfake robocall impersonating President Biden to influence New Hampshire voters is a troubling demonstration of the potential for malicious actors to exploit modern generative AI (GenAI) platforms for electoral manipulation. The increasing accessibility and capabilities of sophisticated GenAI tools such as ChatGPT and Google’s Gemini (formerly Bard) raise concerns about the potential for mass influence campaigns, automated trolling, and the dissemination of deepfake content aimed at disrupting the democratic process in the United States.
FBI Director Christopher Wray has publicly expressed worries about the use of deepfakes for information warfare in the upcoming presidential campaign, highlighting the risk that state-backed actors may employ these deceptive techniques to influence geopolitical balances. Furthermore, the automated nature of GenAI presents the risk of facilitating coordinated inauthentic behavior networks that seek to disseminate disinformation through fake news outlets and social media profiles, ultimately undermining public trust in the electoral process.
Padraic O’Reilly, the chief innovation officer for CyberSaint, acknowledges the substantial risk posed by the rapid evolution of GenAI technology. He emphasizes the potential for microtargeting with AI-generated content on social media platforms, which could lead to the proliferation of highly convincing disinformation with limited bot-like characteristics that could appear authentic to unsuspecting individuals. O’Reilly raises the alarm about the ominous impact of marrying stolen targeting data with AI-generated content, citing previous instances of deepfake content featuring political figures like President Biden and Elizabeth Warren.
The convergence of social media and readily available deepfake technology presents a formidable challenge to democratic traditions and information-sharing, as the potential for widespread dissemination of tailored disinformation threatens to further divide an already polarized citizenry. O’Reilly warns of the emergence of alternative realities fueled by AI-generated disinformation, highlighting the urgent need for proactive measures to address the dangers posed by these developments.
Amid these concerns, experts emphasize the imperative for election officials and campaign teams to be vigilant and proactive in defending against AI-powered threats. James Turgal, vice president of cyber-risk at Optiv, stresses the importance of implementing robust controls over content and providing comprehensive training to volunteers and workers to safeguard against the risks posed by AI-powered deepfakes and social engineering attacks. Additionally, the need for regulations encompassing watermarking for audio and video deepfakes is recognized as a crucial long-term strategy to combat the growing threat posed by GenAI technology.
The Federal Communications Commission (FCC) has taken steps to address the issue by declaring AI-generated voice calls as “artificial” under the Telephone Consumer Protection Act (TCPA), thereby prohibiting the use of voice cloning technology and providing state attorneys general with new tools to combat fraudulent activities. However, the rapidly evolving nature of AI technology underscores the need for continued vigilance and adaptability in developing effective regulations and safeguards to mitigate the risks associated with GenAI.
The potentially transformative impact of GenAI on the electoral landscape underscores the urgent need for comprehensive strategies combining regulatory measures, robust cybersecurity protocols, and enhanced public awareness to safeguard the integrity of democratic processes in the face of evolving technological threats.