Study Shows AI Is Easily Capable of Generating False Election Information Using the Voices of Prominent Political Figures

A digital civil rights group stated on Friday that as high-stakes elections in the U.S. and European Union approach, publicly accessible artificial intelligence tools can easily be used to create convincing election lies in the voices of prominent political figures.

The Center for Countering Digital Hate, based in Washington, D.C., tested six popular AI voice-cloning tools to see if they could generate audio clips of false statements about elections in the voices of American and European politicians. The results showed that out of 240 tests, the tools successfully created convincing voice clones 80% of the time. These clones included fake statements from figures like President Joe Biden and French President Emmanuel Macron.

The research revealed a significant gap in safeguards against the misuse of AI-generated audio for misleading voters. With the technology becoming more advanced and accessible, experts are increasingly concerned about the potential for election disinformation. While some tools have certain restrictions in place to prevent the generation of false information, the study found that many of these barriers were easily bypassed.

Only one of the companies behind the tools responded to inquiries about the research. ElevenLabs stated that they are continuously working to improve their safeguards.

The lack of regulations to prevent the abuse of these AI tools leaves voters vulnerable to deception during important democratic elections. With major elections happening in the U.S. and E.U., the risk of AI-generated disinformation spreading is a cause for concern.

The Center for Countering Digital Hate conducted the research using Semrush to identify the most popular AI voice-cloning tools. Real audio clips of politicians were used to prompt the tools to create fake statements, including warnings to voters and confessions of election manipulation.

Overall, the study showed that these AI voice-cloning tools lacked sufficient safety measures to prevent the misuse of politicians’ voices or the production of election disinformation. Some tools had barriers in place, but these were easily circumvented.

While some platforms are working to enhance their safeguards, there is a pressing need for more regulation and oversight to prevent the spread of AI-generated deception.

The findings underscore the urgent need for action to address the threat of AI-generated disinformation in elections. Lawmakers and technology companies must work together to establish standards and regulations to protect the integrity of democratic processes.