Rising Threat of Audio Deepfakes in Political Arena: The Case of a Fake Biden Robocall

Washington, D.C., is abuzz with concerns about the integrity of upcoming U.S. elections as artificial intelligence (AI) creates new avenues for disinformation. A robocall, imitating President Joe Biden's voice, told New Hampshire residents to abstain from voting in the Democratic primary, showcasing the potential for misuse of deepfake technology in electoral processes.

The unsettling robocall, which featured a distinctly Biden-esque "What a bunch of malarkey," has fueled apprehensions about the future of election security in the wake of these technological advancements. Prompted by the incident, New Hampshire authorities are investigating what could be a case of voter suppression, and campaigners are calling for stringent controls on AI generative tools—or perhaps a complete prohibition on robocalls.

Cybersecurity experts predict that audio deepfakes could become a significant factor in the 2024 presidential race, as tools that facilitate voice cloning become more accessible, affordable, and difficult to trace back to their source. Pindrop, a cybersecurity firm at the forefront of researching this issue, pointedly states that we are merely seeing "the tip of the iceberg."

Pindrop has identified ElevenLabs, an AI voice cloning startup, as the tool used to create the deceptive robocall impersonating President Biden. Despite the company's professed "safety guidelines," which allow cloning of political figures' voices for parodic purposes, the misuse of this technology for political machinations is a stark reality.

As AI-generated robocalls come under scrutiny, US regulators are considering taking legal action to declare the technology illegal, given the real-world implications of their potential misuse. Public advocacy groups stress urgency in policy-making to avoid "electoral chaos" due to these deceptively convincing deepfakes.

These fears are grounded in the broader impact of AI technologies, which can fabricate hyperrealistic videos, text, and, most notably, audio. The impersonation of public figures such as celebrities and politicians has raised global concerns, particularly in audio form, due to the difficulty of discernment and verification.

The complexity of generating and disseminating fake audio content adds to the polarization of political discourse, gnawing away at trust in media and allowing disclaimers of authenticity on factual evidence. China's powerhouse ByteDance has already introduced StreamVoice, a technology that transforms one's voice in real-time, further emphasizing the wide-reaching implications of voice cloning technology.

Experts suggest that audio watermarks or digital signatures could create a layer of security and authenticity for genuine content, but effective regulation remains a challenge. There is a pressing need for investment in trust and safety measures, as well as a commitment to maintaining election integrity in the age of advanced AI.

In conclusion, the fake Biden robocall serves as a cautionary tale about AI-driven electoral disinformation. Policymakers, tech companies, and civil society must collaborate to forge an effective response to safeguard democratic processes against the insidious reach of deepfakes.