- OpenAI says it recognizes the risks of using AI-generated voice tools.
- The AI company has revealed details of its Voice Engine tool, which has not yet been fully released.
- The company listed security measures but did not say whether the model would be widely released.
OpenAI recognizes that AI-generated voice tools can be a sketchy business.
In a blog post sharing results from the early testing stages of a new synthetic voice tool, the artificial intelligence company cited concerns about using AI to replicate human voices, especially in an election year.
OpenAI’s Voice Engine tool, which the company says it first developed in late 2022, uses 15-second audio clips of real people’s voices to create eerily realistic, human-like voices. Create a replica of.
And users can say anything to that voice, even in other languages.
The tool is not yet publicly available, and OpenAI says it is still evaluating “whether and how to deploy this technology at scale.”
“We recognize that there are significant risks in generating audio that resembles people’s voices, which is of paramount concern, especially during an election year,” OpenAI said in a blog post. “We are working with U.S. and international partners in government, media, entertainment, education, and civil society to ensure we include their feedback as we build.”
OpenAI is currently using this tool to power ChatGPT’s “read-to-speech” functionality and the company’s text-to-speech API.
Late last year, OpenAI began expanding the tool externally, working with a “small group of trusted partners” to test the Voice Engine for purposes such as children’s educational materials, language translation, and medical voice recovery. said in the post.
OpenAI requires partner organizations to follow strict policies in order to use the voice engine, including obtaining consent from all individuals to be impersonated and notifying listeners that the voice is generated by AI. He emphasized that there is.
“Due to the potential for malicious use of synthesized speech, we are taking a cautious and informed approach toward widespread release,” the company said in a statement. “We want to start a conversation about the responsible introduction of synthetic speech and how society can adapt to these new capabilities.”
The company said it did not yet know whether it would make the tool publicly available, but urged policymakers and developers to take steps to prevent dangerous misuse of the technology it is developing.
For example, OpenAI proposed establishing a “banned audio list” to prevent the reproduction of prominent voices such as politicians and celebrities without their consent.
The company also recommended that banks stop using voice-based security authentication and that researchers develop technology to track whether voices are real or fake.
Axel Springer, Business Insider’s parent company, has a global deal that allows OpenAI to train models based on its media brands’ reporting.
Axel Springer, Business Insider’s parent company, has a global deal that allows OpenAI to train models based on its media brands’ reporting.
