By Consultants Review Team
For the second time in as many months, OpenAI has had to explain its text-to-audio tool, reminding everyone that it is not yet publicly available and may never be.
"It's critical that people all over the world understand where this technology is going, whether we eventually deploy it widely ourselves or not," the business said in a statement put on its website on Friday. "That is why we want to explain how the model works, how we utilize it for research and education, and how we are adopting safety measures based on it.
Late last year, OpenAI released its Voice Engine, which uses text inputs and 15-second audio snippets of human voices to "generate natural-sounding speech that closely resembles the original speaker," to a select set of people outside the firm. The application can generate genuinely human-sounding voices in a variety of languages.
At the time, the business stated that it chose to preview the technology rather than fully distribute it in order to "bolster societal resilience" in the face of "ever more convincing generative models."
As part of those efforts, OpenAI stated that it was actively working on phasing out voice-based authentication for bank account access, exploring policies to protect the use of individual voices in AI, educating the public on the risks of AI, and accelerating development of audiovisual content tracking to ensure users know whether they are interacting with real or synthetic content.
Despite these attempts, distrust of technology endures.
Bruce Reed, President Joe Biden's AI chief, once admitted that voice cloning is the one thing that keeps him awake at night. In March, the Federal Trade Commission stated that fraudsters were utilizing AI to enhance their job, including voice cloning techniques that made it difficult to discern between AI-generated voices and genuine ones.
OpenAI tried to soothe such concerns in an amended statement released on Friday.
"We continue to engage with US and international partners from across government, media, entertainment, education, civil society, and beyond to ensure we are incorporating their feedback as we build," the business stated.
It also stated that whenever Voice Engine is updated to its current model, GPT4o, it will offer additional vulnerabilities. The business stated that it is "actively red-teaming GPT-4o to identify and address both known and unforeseen risks across various fields such as social psychology, bias and fairness, and misinformation."
The greater concern, of course, is what will happen once the technology is widely available. Furthermore, it appears like OpenAI is bracing itself.
We use cookies to ensure you get the best experience on our website. Read more...