Meta announces new AI model that can replicate the voices of loved ones, but says it's too risky to release

• Meta has announced a new generative AI model that can replicate voices.
• The company said the model could speak six languages.

Meta has announced a new generative AI model that can replicate voices.

The company said the product, which is called Voicebox, can speak six languages. In a news release posted last week, Meta hailed the model as a "breakthrough in generative AI for speech."

Meta said it wasn't publically releasing the model, however, citing the potential risks of the technology.

"There are many exciting use cases for generative speech models, but because of the potential risks of misuse, we are not making the Voicebox model or code publicly available at this time," the company said in a research post. "While we believe it is important to be open with the AI community and to share our research to advance the state of the art in AI, it's also necessary to strike the right balance between openness with responsibility."

Representatives for Meta did not immediately respond to Insider's request for comment, made outside normal working hours.

Meta said in the news release that the model could allow visually impaired people to hear messages from friends in their voices or allow users to speak in foreign languages in their own voice. The company also said the tech opened up the possibility for creators to edit audio tracks for video or create more natural-sounding voices for virtual assistants.

The company said it was looking forward to "seeing how other researchers build on our work."

Meta has been keen to emphasize its AI-development plans. The Big Tech company has been vocal about its ambitions in the space, particularly around building AI into its products. This is notable, given that its lack of consumer-facing AI products has made it less visible.