After DeepSeek, the tech world is shocked by China's new AI
TikTok platform owner from China, ByteDance, has surprised the technology world again with its new AI product.
The Chinese tech giant has introduced a new artificial intelligence (AI) system called OmniHuman-1. The remarkable point of this AI system is that it has the ability to create fake videos (deepfake) with a realism that is almost indistinguishable from real videos.
More amazingly, OmniHuman-1 can export a video with just a photo and a voice recording. Users can customize elements like aspect ratio and cropping. Notably, this AI is also capable of changing body movements and gestures in the video.
While ByteDance has touted the authenticity of OmniHuman-1's deepfake videos, experts have noted that the results aren't always perfect, with some poses looking a little odd. For example, in a fake Albert Einstein lecture like the one below, the fake face is nearly perfect, but the hand movements are unnatural.
To develop OmniHuman-1, ByteDance reportedly 'watched' 18,700 hours of video data, using an 'omni-conditions' method that allows the model to learn from multiple input sources such as text, audio, and body posture at the same time. The researchers said this method helped their AI significantly reduce data loss.
Despite its impressive achievements, the tech world is concerned about OmniHuman-1 because deepfake technology is causing a lot of problems. Recently, it has been used to spread misinformation and fraud, especially in the context of the 2024 US election campaign. There have been many cases where deepfake audio and video recordings have been used to deceive voters.
Last year, deepfake scams cost people billions of dollars, including one in which a fraudster used AI to impersonate actor Brad Pitt to trick a woman into sending $850,000 under the pretext of 'medical treatment'.
Advertisement
Advertisement
A product created by OmniHuman-1.
Hundreds of AI ethics experts are now calling for regulations to govern deepfake technology. ByteDance may not release OmniHuman-1 due to its potential consequences, but the development of AI in all areas, including crime, seems unstoppable.
SEE MORE PERIODS
1 . 165166167168
First periodLatest period
Saved post successfully
You can review saved articles on the Saved Articles page.
Agree
You should read it
- Even Deepfake fraud detection tools can be deceived
- Deepfake AI scam wave: Exploiting billionaire Elon Musk
- Deepfake technology has the potential to disrupt financial markets
- What is Deepfake?
- Free online artificial intelligence (AI) course of Finnish university
- Learn about Artificial Intelligence (AI)
- How to Spot Deepfake Videos
- Samsung's Deepfake could make dark monk Rasputin sing like a real thing
- Online courses on artificial intelligence (AI), certification
- 91% of technology managers believe that AI will be the center of the next technology revolution
- McAfee expert explained how deepfake and AI are drilling through the cyber security wall
- [Infographic] Benefits and hazards from Artificial Intelligence
May be interested
Apple is about to surprise users with iOS 18.3.1
One UI 7 still can't be deployed because of Samsung's mistake
Americans are reluctant to buy new iPhones because of TikTok
Having struggled in many countries, why is TikTok still popular in Vietnam?
Meta is about to have a big layoff in many countries
YouTube rakes in advertising revenue, despite user frustration