Meta has unveiled a new artificial intelligence (AI) tool named ‘Voicebox’ that is capable of generating speech in six different European languages. However, the company has decided not to release it to the public yet due to the potential for unintended harm. While Voicebox outperforms other speech-generation AIs in most areas, its ability to create reasonably accurate text-to-speech replications of a person’s voice with only a two-second audio sample holds a significant risk of abuse in the wrong hands.
In the past, people have already misused similar tools to create fake revenge porn, but Voicebox could take things to another level entirely. For instance, nefarious actors could easily produce highly realistic replications of the voices of public figures like politicians to deceive people and further their agendas, possibly inciting hate speech or even starting a war.
Although similar tools already exist in the market, none are as capable as Voicebox. It is simply too powerful and able to almost perfectly replicate a person’s voice, which is why Meta deemed it too dangerous to release. Instead, they are publishing a research paper outlining its capabilities and have developed a classifier tool that can identify Voicebox-generated speech from actual human speech.
Nevertheless, AI speech generation tools may provide significant benefits in the future, such as providing naturalistic speech to people who are mute or unable to communicate. It could also perform real-time translation, making effective communication possible despite language barriers. There are numerous other applications for this technology, such as improving and editing recorded speech.
Meta’s cautious and considered approach is a welcome sight in a world where AI technology is increasingly prevalent. Other companies, including Microsoft, have shown a lack of caution in unleashing new AI products, sometimes resulting in negative unintended consequences. However, Meta’s decision to delay the release of Voicebox shows that they are taking the potential for harm seriously.
It remains to be seen how long Meta will hold off on releasing Voicebox to the public, but for now, they are prioritizing safety. AI technology is advancing rapidly, and while it has the potential to benefit humanity greatly, it is also vital to consider the risks. Meta’s careful and responsible approach to this technology is a positive sign for the future.
