HomeEconomyOpenAI's Text-to-Speech Model Raises Deepfake Concerns, Halts Wider Rollout

OpenAI’s Text-to-Speech Model Raises Deepfake Concerns, Halts Wider Rollout

OpenAI has showcased its early results from a trial of a text-to-speech model capable of generating human-like voices, underscoring both the advancements in artificial intelligence and the potential risks associated with deepfakes.

The company, which had briefed reporters earlier this month on the feature, is sharing preliminary demos and use cases from a limited preview of the text-to-speech model named Voice Engine. Initially intended for broader release to around 100 developers, OpenAI opted to scale back after receiving feedback from various stakeholders, including policymakers and industry experts.

“We recognize that generating speech resembling people’s voices carries serious risks, particularly in an election year,” OpenAI stated in a blog post. “We are actively engaging with stakeholders to ensure their input shapes our development process.”

Unlike previous audio generation efforts, Voice Engine can replicate individual voices with specific cadences and intonations using just 15 seconds of recorded audio. During a demonstration, a sample of OpenAI CEO Sam Altman’s voice sounded indistinguishable from his actual speech, raising concerns about the potential misuse of such technology.

OpenAI’s partners, including the Norman Prince Neurosciences Institute at Lifespan, are exploring applications such as voice restoration for patients. Spotify has also piloted the technology to translate podcasts, highlighting its potential in various industries.

In response to concerns, OpenAI is implementing safeguards such as usage policies, consent requirements from original speakers, and disclosure protocols for AI-generated content. Additionally, the company is soliciting feedback from experts before considering a broader release.

The preview serves as a reminder of the need to address societal challenges posed by advanced AI technologies, urging institutions to reevaluate security measures like voice authentication and prioritize public education on detecting deceptive AI content.

Read Now:Salon Hair-Straightening Treatment Leads to Kidney Damage: Study

[responsivevoice_button buttontext="Listen This Post" voice="Hindi Female"]

LEAVE A REPLY

Please enter your comment!
Please enter your name here

RELATED ARTICLES

Trending News

ISRO’s Satellites Capture Maha Kumbh Mela From Space, Showcasing World’s Largest Religious Gathering

Prayagraj: The Indian Space Research Organisation (ISRO) has used its advanced satellite technology to beam down awe-inspiring images of...

Global Economy Faces 50% GDP Loss by 2090 Without Climate Action, Warns Report

A stark warning has been issued about the economic and humanitarian costs of inaction on the climate crisis. According...

Delhi Police Seizes ₹47 Lakh in Cash Amid Election Code Enforcement

In a significant move during the enforcement of the Model Code of Conduct (MCC) ahead of the upcoming Assembly...

India and France Unite to Tackle Maritime Threats in Indian Ocean

In a significant move to bolster maritime security, India and France have agreed to jointly assess and address threats...