ETH researchers develop chip for detecting deepfakes

(CONNECT) Researchers at ETH have developed a sensor technology that provides image, video or audio data with a kind of digital fingerprint as soon as it is recorded. This cryptographic signature can be stored in a public register, such as a blockchain. According to a press release, this makes it possible to check at any time whether data is genuine or has been subsequently changed. The aim is to make it easier to recognize manipulated content such as deepfakes.

Any subsequent changes to the files leave traces, as they no longer match the original signature. "Trust in digital content is dwindling. We wanted to develop a technology that allows people to check whether something is genuine," says Felix Franke, who co-developed the chip at ETH.

In order to manipulate data, the chip would have to be physically attacked. "A technical effort that would be so high that the mass production of manipulated content for social media platforms would be practically impossible," says co-developer Fernando Cardes.

The technology can be integrated into cameras and other sensors. The developed chip is currently still a prototype. Further development steps are necessary for widespread use. However, the researchers see potential for commercial use and have already submitted a patent application. ce/as