Great collaboration with the group of Andreas Hierlemann at D-BSSE, ETH Zürich (bsse.ethz.ch/people/detai...) and Sebastian Bürgel now at Gnosis.

This would require a broader ecosystem — sensor manufacturers, trusted public infrastructure, and platforms that actually perform verification.

But the technical foundation is possible.

We cannot stop synthetic media from existing. But we can build systems that make authentic recordings provable.

Of course, this raises important privacy questions.
But such systems could be designed in ways that preserve user privacy rather than exposing everything someone records.
There’s more discussion on this in the paper and supplement.

The goal is simple:

No tampering with digital data without a trace.
No source data without a source sensor.

C2PA | Providing Origins of Media Content Enhance digital safety through the use of content authenticity tools. C2PA provides a way to ensure content transparency by analyzing the origin of media.

Combine that with standards such as C2PA (c2pa.org), and later edits could also be tracked transparently.

Add a secure clock, location (GPS), and social media like bluesky that does the verification directly on upload, and you could get a green check mark next to each authentic recording.

You get a photo.
You check its signature.
You check that the corresponding public key belongs to a trusted manufacturer.
You verify that the signature matches both the key and the image.

If all checks pass, you have strong evidence that the image came from a real physical sensor.

How would that work in practice?

The sensor’s public key would also need to be stored in a trusted public repository, for example by the sensor manufacturer.

Then verification becomes possible for anyone.

So if you later receive an image or recording from such a sensor, you could verify for yourself that it really originated from that physical device — and was not generated or altered somewhere downstream.

How the technology works: a real-world event (1) is recorded by a camera whose sensor chip generates both the image data and a cryptographic signature at the moment of capture (2). Once stored in a public register (3), the signature can later be used to verify that the recording is authentic and has not been altered (4). (Graphic created using AI: Felix Franke / ETH Zurich)

The sensor contains a cryptographic circuit that signs the recorded data with a private key that exists only inside the hardware.

That signature is then published to a trusted, immutable public register.

A blockchain is one possible way to do this.

In the paper, we built a prototype sensor designed to do exactly that.

The basic idea: add cryptography directly inside the sensor, at the point where the physical signal becomes digital data.

Can a digital recording be made trustworthy again?
Can we link the act of recording something real directly to the digital data that is produced?

We think the answer is yes.

A physical artifact in a museum — a sword, a manuscript, a fossil — is evidence of its own existence and history.

A digital recording is different. A photo or video can now be generated artificially, without any corresponding real-world event.

In-sensor cryptographic signature generation to link a physical process and an immutable digital entity - Nature Electronics A monolithic cryptographic sensor, which integrates a sensor device with a hashing unit and a cryptographic signature unit, can be used to create a recording system that provides a cryptographic link ...

Our new paper in Nature Electronics is out now
www.nature.com/articles/s41... (paywalled)
rdcu.be/e9JPM (no paywall)
going back to my time at ETH Zürich.

Perhaps the most famous deepfake image to date: the Pope in a stylish outfit. (Image: Pablo Xavier / Midjourney)

Do we have to accept a future awash with fake images, videos, and recordings — one where trust in media and online communication steadily erodes?

We think the answer is no.

Technology can be used to undermine trust. But it can also be used to defend it.

🤖As deepfakes rise, a chip could bring back trust

A new sensor technology secures data at the moment it is captured, making manipulation detectable.

Image © Felix Franke, IOB, 2026, generated with AI

🧵👇

Physiological basis of resolution acuity in vision - Nature Communications Neural mechanisms underlying high visual acuity are not fully understood. Here the authors show that high resolution visual information is transmitted from the retina to the brain by neurons in t...

Most vision scientists would bet that single-cone receptive fields exist in primate fovea, but it was not proven until now.
It took years of sweat and $ to combine #AOSLO and electrophysiology. Congrats to PhD student Keaton Ramsey and his mentor Lawrence Sincich.

www.nature.com/articles/s41...

In vivo single-cell gene editing using RNA electroporation reveals sequential adaptation of cortical neurons to excitatory-inhibitory imbalance The balance between excitatory and inhibitory neurotransmission is fundamental for normal brain function, yet the adaptation of individual neurons to disrupted excitatory-inhibitory balance is not wel...

Gene editing of single, targeted neurons in vivo is now feasible. We are proud to present our preprint for highly efficient single-cell electroporation using RNA. With @alex-fratzl.bsky.social, @munzlab.bsky.social, Botond Roska @iobswiss.bsky.social
#neuroskyence
www.biorxiv.org/content/10.1...

Thrilled to share that our work is now published in Science! ✨

We found a preference for visual objects in the mouse spatial navigation system where they dynamically refine head-direction coding. In short, objects boost our inner compass! 🧭

www.science.org/doi/10.1126/...

🧵1/

👁🧠 NEW: The human retina has its own built-in timing fix!
A Nature Neuroscience study led by @ethz.ch | BEL Lab, IOB Basel, @unibas.ch shows that photoreceptor signals – even those that travel farther – arrive in sync! >https://u.ethz.ch/S4DvU
#Retina #Neuroscience #Bioengineering #ETHZurich

The answer: Photoreceptors need to connect to other neurons in your eye (via yet other neurons) and those can make the connection to your brain.

These neurons are called retinal ganglion cells, the one you see in the image.

But why a ring?

Why does it look like a ring?

Because it is a ring. In the very center of your eye, you only have photoreceptors to detect light!

But photoreceptors do not connect to your brain.

So how to get the information from the photoreceptors out?

You are currently using them to read this text!

The image basically shows how visual information about this letter here, yes this one (and this one, too!) is traveling from your eye to your brain.