Zero Knowledge, Full Control by Denis Jaromil Rojo // Web3Privacy Now Rome Meetup 2026

Um so I I I assume we are a lot of cryptographers in this room. I will not ask to raise your hand. It's a zero knowledge assumption. Um Yeah, I'm also working as applied cryptographer since 12 years now. It's a good life, especially considering I was in Bitcoin in 2011.

So I've seen all the crypto shuffle and um I'm now uh fast forward. I just became the the the scientific director of Plan B Lugano. Lugano is the Bitcoin city in Europe. Full of Bitcoin maxis. I wonder how I made it.

I'm not wearing a MAGA hat, you know. But but uh no, it it it it's working. It's working well because there are people also among Bitcoiners you can looking for humanity in this in this uh context. And uh because people are starting to go beyond the football uh attitude on which team you're playing on and and looking for values and for long term. I'd say we are on the plateau.

We passed the hype. Now AI is the hype. So we can reason a little bit better among us. I've seen this hype starting. I was in cryptography slightly before the hype started with Bitcoin.

So it it became incredible the space. I mean, the community dissolved completely. I'm now reconnecting with some people that I remembered at the the very beginning. Uh some of you also. And um Yeah, what I did in meantime was a skip most of the crypto conferences in the past 12 years and come to web three privacy.

So, thanks for building this because it feels human and feels like we can talk about tech and politics. Thanks, Sterling. And I would like to to start from really what you say that the metadata I will get there. I think it's very important that you um that you draw the picture of the kill chain, no? What What is being used the metadata for?

And there is a gradient of things it's being used for very unethical purposes uh without even killing people. Uh framing them, silencing them, um mobbing in various ways. So, uh what I'm going to say uh I would like to share a a a um reflection on on ZK, this tech, and what is what it what could become. Uh I say it because um I think that on the long term we will win. So, what you said is also clear.

You You talk about reasonable values. No, I mean, yes, we are activists, but at the end we are relating with with a large mass of people. And the people out there, the people chased by ICE, they are the majority the in the US or or the people here like reasoning with you know, with with the fact that in a neighborhood you need a social center where together, where to make a a concert, where to have a social space where you can enter without paying enormous amounts. This is reasonable for the people living in cities, in dense settlements, and for most of us. Unless you become a you know, sociopathic billionaire in the crypto shuffle.

Even then you will save yourself. Um so, I think we will win on the long run, and I think those of us who are reasoning of what they do or what we do with our career, with what we know how to do, uh may want to choose the long-term winners and not just like the the easy win, the quick win. Because we all put our names on our papers and what we do. So, what is the danger today that I see working as a cryptographer in zero knowledge? I met it in my own work.

My journey was through um the European digital identity. Before that, in 2012, we did the first uh blockchain research project ICT 12B funded by DG Connect, the same department, the same uh directory general in the European Commission that is doing today the European digital identity. Uh we did a project for 3 years. It became a flagship project. George Danezis was there.

Other colleagues. Um Coconut was born, then became what is today running Nym as well by Alberto Sonnino and George. Um Libra uh we hired four of us. The commission didn't understood what the hell was happening, why Facebook wanted four of our scientists on their side. Um It was BLS based, so it's not post-quantum today, but it it's fun.

It's a fun uh encryption scheme that you may know. Um so, what happened what happened in between was that the values we were that that the value proposition, not the values, the value proposition that we were we were proposing for zero knowledge use in the presentation of documents was understood halfway. There was a business proposition uh many use cases, age verification, students, tourism, uh hotels, saving uh data people, what not. But the execution today presents many problems. They are given by the fact that people pass hand by hand like, you know, someone talks to someone else, someone talks to someone else.

Not everyone is really aware of what is happening. A lot of engineers in the room, very few social researchers, very very few people that from a interdisciplinary point of view can analyze really what is becoming this technology. And today we are faced with this huge Trojan horse which is age verification online. Some states in the US even made laws for that. So it's being built into the operating system.

You may know I created the fork of Debian called that one, removing systemd. Well, 2014 we were right. Systemd is building birth date records into the operating system to handshake with the web and let you navigate online. So, what are we creating? This is a equivalent of the Italian uh cigarette vending machine where you have to input all your data.

So, in Italy, if you want to buy a package of cigarettes, if you unluckily for you are a smoker, then you will go and buy a cigarette, you need an Italian uh codice fiscale is called. So, actually it would be the BSN in the Netherlands, it would be the the social fiscal code, the identifier for the state and the social insurance and the health insurance, one identifiers uh with your name, birth date and this identifier in a card that you input into a machine selling cigarettes. Any that you see on the street. So, it could be skinned, whatever. You insert it and you to get your cigarettes.

Obviously, you see a lot of people asking, "Can you give me that?" And those of you who are not Italians or luckily holding such a pass will have to ask this to buy it if you're late at night and all the tobacco closed. So, we are doing the same to the internet. We're doing the same. Like we are we are building the first global control access layer on top of the internet.

You have to say, "How old are you?" And we are mandating to platforms that serve things that in some legislations are deemed to be adult only. Now, let alone the problem of legislations may be different. Some people believe that porn can be consumed at 16 or 18 or whatnot. We are mandating those platforms to process this data.

And perhaps we are going to explain them, "Oh, you can use zero knowledge. This is the prover. This is how you run it. Okay, good luck." And um and we'll say, "Yeah, this these platforms will not gain a knowledge of the data of the people because it is a zero knowledge presentation.

We'll just be sure that people are above a certain age." So, this is a Trojan horse for mass surveillance, for metadata collection, and it's also not helping us to serve people with an with with a technology that help them. Because yes, zero knowledge can help, but zero knowledge need to be computed. And so here I get to my point on which side we are want to be because we are on a slippery slope. If we put zero knowledge to be computed into a platform that can access the data at input before it's processed, then we are giving this platform knowledge of the data.

Another assumption. If we let this computational platform gather telemetrics on all what it does, which operating systems on mobile phones do, and bind it to the act of presentation, then we are also informing this this platform with the metadata of this presentation and who does it. So, my hypothesis is that even without knowing the input, which an operating system can access at kernel level anyway, even without knowing the input to a zero knowledge machine, the operating system manufacturer can use all the telemetry and all the data that is already gathered about the user plus the information of a presentation at a certain date and to a certain relying party, also known as verifier. This is a hell of a lot of metadata. Even without knowing the input, which at kernel level both Android or uh iOS could access.

So, we are being used and abused as cryptographers when we are asked to put zero knowledge into a state system that wants to grant people privacy and this zero knowledge computation will run into a platform that will acquire even more information about the people. To the policy makers, I would say, "You are being abused." Because these platforms, the operating system manufacturers, will know more about the citizens than you know. If you're running a state, I imagine myself running a a small city like Lugano that I can relate better. We are 70,000 inhabitants and it's Switzerland, so it's all federated.

Um you know a hell of a lot about your citizens. I have seen the Department of Statistics is frightening and amazing at the at the same time. You know also with the police department all the presentations of every time someone is asked of a driving license. All this information will come out of the hands of the state in the hands of the very OS manufacturers of the devices that are processing. What will happen the next day?

The operating system manufacturers will go to the policy makers and say, "Would you like a new dashboard? I have it. That's the cost. We can provide it. It's all new."

And there you go. It's the Palantir business model. We are being abused as cryptographers this way. The same way um I I also analyze in my in my um in my dissertation how how deep um deep machine learning was abused in many cases in many legal cases um of crimes committed by what we call today AI. Uh systems like ADAS ADAS used by the police, uh systems like other other acronyms in the Netherlands used to assess subsidies that single mothers uh needed to receive in certain parts of the Dutch state state.

They all failed calculations. So, these were crimes committed by AIs, what we call AIs today. These crimes in court were defended often by the fact that oh, it's a deep learning widgets. It's a deep learning machine. It's like so deep technology, we don't know what is happening inside.

It's funny this word how it was used, deep. It's like so deep that we cannot explain it anymore. It's non-deterministic. It's blaming the machine instead of my own responsibility of deploying such a system at large scale. We are being used as cryptographers today doing ZK.

If we do put ZK into a system of mass surveillance, ignoring the fact that the integration details will actually erode people's privacy. That's why today I'm proposing to the city I work for to use cards, smart cards. That's why today I would be very interested to talk to any of you who know a way to make it cheap and easy to produce small chips, identity cards that can compute zero knowledge, but are not embedded into a system of mass surveillance like a mobile operating system. I know it's hard. I know it doesn't exist, but it's just an I'm starting to think about it.

Or any of you who have an idea how provable isolation can be done in an operating system so it can be proven that the computation of ZK is is isolated. But we would have a very difficult case there to isolate it from any possible, you know, activation of that part and then telemetry uh attached. So, isolation in that case, I think it wouldn't work. And um yeah, this is this is where I'm at today. This is what I wanted to share with you and I'm up for discussions at the cocktail.

Cheers.