With the growing popularity of mainstream AI tools on the cusp of the 2024 US presidential election, AI-generated disinformation is not just a fear – it is a reality. On January 22, the Recent Hampshire Department of Justice released a statement that the people had been sentenced recorded deepfake audio of Joe Biden, telling them not to vote in the state’s primary elections. The appeal encouraged voters to “save” their vote, falsely noting that “your vote matters in November, not this Tuesday.”
A few days later, the AI launches ElevenLabs suspended the creator of the fraudulent Biden recording– Bloomberg reported.
Eleven laboratories is a model-powered AI voice generator that, according to its website, can add human-like inflections to a voice based on context. The generator does thousands of ready-made AI voices to choose from or you can create your own. Bloomberg reported that Pindrop Security Inc. voice fraud detection company discovered that the automated AI Biden call was made using ElevenLabs.
“Our goal is to prevent the misuse of AI audio tools, and we take any misuse extremely seriously,” ElevenLabs told Bloomberg. ElevenLabs website stipulates that deepfakes against politicians can only be used in certain cases, including caricature, parody or satire. When the company learned of Biden’s deepfake, it investigated and suspended the account responsible, a source told Bloomberg.
In an interview with The Hill, computer science professor at Carnegie Mellon University Kathleen Carley said: Biden Robotcall is the ‘tip of the iceberg’ regarding voter suppression attempts. Carley added that this is a sign of what may be to come.
ChatGPT developer OpenAI is already trying to quash disinformation by publishing plans to protect election integrity. Shortly thereafter, the company suspended a programmer who created a bot for the longtime Democratic candidate.
As such, we must remain vigilant about what we see – and hear – this election season. As Mashable technology reporter Cecily Mauran warned: “The idea of an Internet dominated by AI-generated content is already emerging, and it doesn’t look good.”
