Microsoft teases deepfake AI that's too good to release

Microsoft has developed VASA-1, an advanced deepfake AI framework capable of creating convincing videos from a single image and an audio sample. Despite its potential, they won't release it due to the high risk of misuse, aligning with growing concerns and regulatory efforts surrounding AI impersonation and deepfakes.

• VASA-1 can animate still images to speak.
• Microsoft prioritizes ethical use.
• No plans for public API or product.
• Deepfake tech can be misused for fraud.
• Regulations are emerging globally.