Archive — US — techcrunch
A year later, OpenAI still hasn’t released its voice cloning tool
OpenAI's voice cloning tool, Voice Engine, announced last March, is still in limited preview with no launch date, reflecting concerns over potential misuse and regulatory scrutiny. Despite its promising applications in areas like speech therapy and customer support, the long delay raises questions about OpenAI's commitment to safe AI deployment amidst growing scrutiny over voice cloning abuses.
Key points
- OpenAI announced Voice Engine in March 2023, claiming it could clone voices from just 15 seconds of audio.
- As of March 2024, the tool remains in a limited preview with no public access.
- The delay is partly due to concerns over potential misuse, especially amid increasing abuse of voice cloning technology.
- OpenAI has been testing Voice Engine with trusted partners to ensure safety and utility before a wider release.
- Voice Engine incorporates safety measures, including watermarking and explicit consent requirements from original speakers for usage.
- Despite various applications, including benefits for people with disabilities, rollout timelines have been postponed amidst concerns about the technology's impact on privacy and security.
Why it matters
- Voice cloning technology has become a significant concern in light of rising scams and deepfakes, especially as AI-generated voices were identified among the fastest-growing scams in 2024.
- The delay in launching Voice Engine reflects a broader trend in the AI industry where companies prioritize safety and ethical considerations over market competition.
- This hesitance mirrors past experiences with technology deployment, where rapid releases led to significant ethical ramifications, such as the initiation of regulations in tech design.
The prolonged preview of OpenAI's Voice Engine indicates a cautious approach to AI voice cloning, balancing innovative potential with pressing safety and ethical challenges as society navigates the complexities of synthetic media.