OpenAI Delays Release Of Voice Cloning Tool, Amid Election Concerns
OpenAI concludes it is too risky currently to release its AI voice cloning tool, amid concern at tech’s impact on upcoming elections
AI pioneer OpenAI has decided to delay the general release of its voice cloning tool, in an effort minimise misinformation ahead of important elections around the world.
Voice Engine was first developed in 2022 and an initial version was used for the text-to-speech feature built into ChatGPT.
But now the firm has announced in a blog post that it is taking a “cautious and informed approach to a broader release due to the potential for synthetic voice misuse.”
OpenAI’s Voice Engine can apparently generate a convincing clone of anyone’s voice using just 15 seconds of recorded audio. In February OpenAI had released a tool (Sora) that can create short form videos just from written text instructions.
Elections concern
The potential of the misuse of this tool, with the US Presidential election taking place later this year, is obvious.
There are also a number of other national elections taking place in 2024. For example India’s general election begins in April, and there is also going to be important elections in the United Kingdom and South Africa, as well as European Parliament elections in the summer.
In January this year New Hampshire legal authorities began investigating fake robocalls pretending to originate from US President Joe Biden, in an apparent effort to employ artificial intelligence (AI) to disrupt the state’s primary elections.
Such is the concern that in February twenty of the world’s biggest technology companies, including Amazon, Adobe, Google, Meta, Microsoft, OpenAI, TikTok and X vowed to take measures against the misuse of AI to disrupt elections around the world this year.
This was evidenced in March when Google confirmed that it would restrict the types of election-related queries that users can ask its Gemini chatbot.
And AI is already being exploited by hostile nation states.
Microsoft recently revealed that nation-state hackers from Russia, China, Iran and North Korea, are already utilising large language models such as OpenAI’s ChatGPT, to refine and improve their cyberattacks.
Voice cloner delay
Into this comes the confirmation that OpenAI is delaying the general release of Voice Engine.
“We first developed Voice Engine in late 2022, and have used it to power the preset voices available in the text-to-speech API as well as ChatGPT Voice and Read Aloud,” blogged OpenAI. “At the same time, we are taking a cautious and informed approach to a broader release due to the potential for synthetic voice misuse.”
“We hope to start a dialogue on the responsible deployment of synthetic voices, and how society can adapt to these new capabilities,” it added. “Based on these conversations and the results of these small scale tests, we will make a more informed decision about whether and how to deploy this technology at scale.”
In its post the company shared examples of real-world uses of the technology from various partners who were given access to it to build into their own apps and products.
“Voice Engine is a continuation of our commitment to understand the technical frontier and openly share what is becoming possible with AI,” OpenAI added. “In line with our approach to AI safety and our voluntary commitments, we are choosing to preview but not widely release this technology at this time.”
Last September Spotify said it was testing an AI-powered system that translates podcasts from English into other languages using the speaker’s own voice.
It said it was initially testing the new feature with a “select group” of podcasters including Dax Shepard and Monica Padman, Lex Fridman, Steven Bartlett and Bill Simmons.
The feature used voice translation technology from OpenAI that Spotify said resulted in a “more authentic listening experience” that is “more personal and natural than traditional dubbing”.
