Intel has developed an AI model, FakeCatcher, that detects deepfakes in real time on video.
The world now has its first real-time #deepfake detection platform.
Learn how @ilkedemir and her team use FakeCatcher technology to find the fakes.
— Intel (@intel) November 14, 2022
According to the developers, the system can distinguish a real person from a fake, including by changes in the color of blood vessels. The algorithm reads this information from multiple points on the face and then processes the data with AI.
FakeCatcher’s accuracy stands at 96%, and the response time is in the milliseconds, according to the developers.
The company outlined several potential use cases for the algorithm. Among them are preventing users from uploading malicious deepfake videos to social networks and helping the media prevent the broadcast of manipulated content.
The company also noted the growing threat from high-quality neural deepfakes.
“Deepfake videos are now ubiquitous. You have probably seen them: clips of celebrities doing or saying things that never actually happened,” said Ilke Demir, a senior research scientist at Intel Labs.
There are no details about plans to release FakeCatcher publicly.
In April, tech giants formed an alliance to counter deepfakes.
In June, major IT companies that own online platforms, joined the fight against disinformation in the EU.
Subscribe to ForkLog’s AI news on Telegram: ForkLog AI — all the news from the world of AI!