this is how it clones the human voice. The risks – Time

OpenAI, the artificial intelligence giant that created ChatGPT, presented a system for cloning the voice: “Voice Engine”. The use of this system will be limited to prevent fraud or crimes, such as identity theft. This AI model can reproduce a person’s voice from a 15-second audio sample, OpenAI said based on the results of a small-scale test. “We recognize that the ability to generate people-like voices carries serious risks, which are particularly relevant in this election year. We work with American and international partners in government, media, entertainment, education, civil society and other sectors and will take their comments into account as we develop the model,” the San Francisco-based company said .

OpenAI ensured that it had adopted “a prudent and informed approach”, before a wider diffusion of the new tool “in consideration of the potential misuse of synthetic voices”. The clarification follows a serious political incident that occurred when a consultant for the presidential campaign of a Democratic rival of American President Joe Biden developed a program capable of simulating the identity of the latter, running for re-election. The voice imitating Biden’s urged voters not to vote in the New Hampshire primary. The United States later banned such use of AI-generated cloned voices to counteract political or commercial scams.

OpenAI specified that the partners testing Voice Engine have agreed to rules that imply explicit and informed consent of any person whose voice is cloned and transparency for those who listen to it, who must be clearly informed that those voices are generated by ‘artificial intelligence. «We have adopted», OpenAi assured, «a set of security measures, including a trademark to be able to trace the origin of any sound generated by Voice Engine, and a proactive control of its use». Last October the White House announced the rules and principles to govern the development of AI, including that of transparency.