OpenAI debuts voice cloning tool, but deems it too risky for public release | Technology

From Al Jazeera: 2024-03-31 23:29:20

OpenAI unveils Voice Engine, a tool that can replicate a person’s voice from a 15-second audio sample, but delays public release due to concerns of misuse in an election year. ChatGPT creator emphasizes the need for responsible deployment of synthetic voices. Company seeks feedback from various sectors to address risks.

OpenAI plans to make an informed decision on scaling Voice Engine based on testing and public discourse. Company advocates for measures ensuring consent from the original speaker and preventing creation of voices too similar to public figures. Concerns about AI misuse in elections arise as countries gear up for major elections in 2024.

AI’s impact on elections highlighted with cases of AI-generated speeches in Pakistan’s parliamentary elections and a robocall impersonating US President Joe Biden. OpenAI implements safety measures in Voice Engine testing, including watermarking for traceability and proactive monitoring of usage to prevent misuse.



Read more at Al Jazeera: OpenAI debuts voice cloning tool, but deems it too risky for public release | Technology