Researchers at ETH Zurich say they have built a prototype sensor chip that cryptographically signs data at the moment it is captured, creating a hardware-rooted way to verify whether an image, video or audio file is authentic. The team says the idea is to let a recording carry proof of origin from the instant it is produced, making later tampering or synthetic substitution far easier to detect.
According to ETH, the sensor signs the data directly inside the chip rather than relying on a later software process. That means the signature can show that the file came from a specific device, indicate when it was captured and confirm that it was not altered afterward. The researchers say the signature could be stored in a public immutable ledger such as a blockchain, allowing anyone to compare the original media against the registered signature and check whether the file remains genuine.
The work is still at prototype stage, and ETH says more steps are needed before the system could become a commercial product. Even so, the concept speaks directly to a growing problem for identity, security and trust systems: deepfake detection often starts only after suspect media is already in circulation. A trusted capture chain would move part of that defense to the hardware layer, where the question becomes not whether software can spot a fake but whether the media can prove it came from an authentic capture device in the first place.
The significance lies in where ETH is placing the trust anchor. Rather than trying only to analyze suspect files after the fact, the prototype aims to make authenticity part of the capture event itself. If that approach proves practical at scale, it could eventually influence how cameras, sensors and identity-sensitive devices establish trust in images and recordings before those files ever reach an onboarding flow, a platform or a social feed.
Sources: ETH Zurich
–
By the ID Tech Editorial Team
