IN Brief:
- ETH Zurich researchers have developed a prototype sensor chip that cryptographically signs data at the exact moment of capture.
- The approach is designed to verify authenticity for images, video, and audio, and could be used with public ledgers to confirm provenance later.
- The work moves anti-deepfake technology closer to the sensor itself, where authenticity can be anchored before software manipulation begins.
ETH Zurich has developed a prototype sensor chip that digitally signs image, video, and audio data at the moment of capture, moving authenticity checking down into the hardware layer as synthetic media becomes harder to distinguish by inspection alone.
The principle is straightforward, even if the implementation is not. A cryptographic signature is generated inside the sensor chip alongside the captured data, creating a verifiable link between a physical event and its digital record. If those signatures are later stored in an immutable public register, the source and integrity of footage can be checked after the fact without relying solely on platform trust, metadata, or software-based watermarking.
The work is notable for where it came from. The concept grew out of biosystems research, where the ETH team had been developing highly sensitive chips to measure electrical signals from living cells. That background turned out to be useful when the group began exploring how cryptographic functions could be integrated directly into sensors. The resulting device is still a prototype, but the team has already published the work in Nature Electronics and filed a patent application.
For electronics designers, the broader point is that authenticity is starting to become a sensor-system design question rather than a pure software problem. As generative tools improve, provenance may need to be established at the source, inside cameras and other capture devices, before the data ever reaches a network, platform, or editing pipeline.