FHEthereum: FHE for Ethereum

Transcript

Hi everybody. I'm Yanis. I'm I'm working at the Ethereum Foundation. Super happy to be here. It's my sixth ECC so always a pleasure to present to present here. I'm part of a team called IPTF for institutional privacy task force. It's a It's a team that was created like last summer. It came from the observation that EF discussing more with institutions, we realized that the main blocker was actually privacy. And so we created this team to actually address this problem and listen to all the requirements that this institution have and build a lot of resources about like how can you use Ethereum in a in a private way.

So how do we do this? How do we answer this problem? We have like many initiatives. The main one is a privacy map. So it's a fully open sourced work in the form of a GitHub repository where you can find a world knowledge based knowledge base about like what are the different use cases these institutions have? What are the building blocks that you can use to address this? We try to form solution in the form of approaches and for institutions that would like to actually like build something we list vendors that actually provide the technology that we we list in the in the map.

We also we also doing like some proof of concepts. So here the the goal is to not only like a speak about these building blocks but have like a hands-on approach of them. And so it's it's having like yeah, more like developing an engineering culture in the in the way we can answer like to these institutions. And also we are doing like some coordination with with stakeholders in like that that are on the on the institutional market and as well as doing like workshops for these institutions where we try to dive a little bit more into a privacy on Ethereum today.

So why do we have this initiative? So as I introduced like we observed that the the main blocker today is privacy while it was like regulation few months years ago. Now that the regulation in some jurisdiction is a little bit clearer, we are seeing a big push from these institutions to like switch from private blockchain to like public blockchain and therefore the the main issue that they have like making the switch is privacy. There is also like a lot of overlap with the one trillion security initiative at the EF and also of course like this notion of leverage that if we find institution that resonate with with Ethereum and we we we we

we have success in onboarding them on Ethereum, it's a huge leverage for the for the world world space but as well as the end user that would use like a these solutions. Yeah and basically Ethereum is like is is public by default. So everything everything is is is public for for an observer. And so we are trying to actually build mechanisms to prevent a all these all these unobserver to to see all these public informations.

So um why Ethereum? Like why these institutions now they want to make the switch and like they are interested they have interest in in Ethereum. I'm not going to explain like here but basically it's like they want to access deep liquidity, they want to censorship resistance because these institutions actually do care about censorship resistance because for them it translate to like no counterparty risk. They need like a settlement shelling point like a neutral place where they can come and like trade. And they also need like a composability that that Ethereum offers compared to like a the other models where you have like a full privacy but you you don't really have like a interoperability.

It's like a kind of a walled garden and you also have like a vendors that provide this technology and it's not very open about like what you can do and and you can end up like in a in vendor lock-in. So yeah, it's quite a challenge but that's that's what what we are trying to address. So how do we work? Like what's what's the our day-to-day job? First is talk to institutions, try to find them to meet them where they are. So participating in like a lot of conferences like go there, try to have discussions with them, organizing workshop as well.

And yeah, try to gather as much real requirements as possible. We don't want to speculate anything here. We just want to to listen to like the the actual problems that that they have and try to to to solve them. We So as I said like all these all the all these discussions, they are open sourced. So we anonymize who we are speaking to but then then like in our open open source resources, you will find like a set of use cases that are actually like curated from from these discussions. And we want this like to be like something that the the the world ecosystem can use as okay so and I want to tackle some institutional use cases and then I can come here like see the the set of

requirements and try to shape my solution going into that direction. Um yeah so we also list a lot of patterns. So the methodology we apply here is like a try to think by the first principles, try to think like in terms of of cryptographic patterns like in terms of building blocks and without no not necessarily like an opinionated view on them and then like from these building blocks try to build a more opinionated view about like how to to build a solution which we name approaches. So basically it would be like all right, there is this institution wanting wanting to do like they want to do like this private

derivatives. How can we how can we how can we build a solution from these first principles? We also like want to do more on this on this on this side like more like open specifications that anybody can like use and and build from it. And also like playbooks were in the form of workshop guides for these these institutions to to actually use and and understand more like the space. So just a a sneak peek of the of of the current state of the mapping. So we have like a 52 pattern listed from shielding to to to to TE to FHE.

We have 20 20 use cases and we also list 21 vendors. I think on that space so that's probably like ECC is a good place to call all these vendors to come and like add their protocol if they want to address like some institutional privacy use cases. It's a very open. You can open a PR, explain what your protocol is doing, what you're taking solving as a institutional problems and then we would be happy to to have you added to the to the vendors as well as if you if you if you have like a um a policy policy expertise. We do have like some some jurisdictions artifacts

trying to explain like how how are the different like rules in each jurisdiction and how it translate in terms of build for builders like what are the the thing I need to implement to to address these requirements. Um a recent change as well is like it's making sure that we give an overview of like cross alignment in terms of solutions. So of course not all these not all the the solutions are like on the same line when it comes to like censorship resistant openness, privacy and security. So we are trying it's something that we just added try to like give as well for institutions like a like some some criteria that they can

use to understand better like guarantees that a solution can offer. So this is like a a breakdown from the privacy cluster of of of Ethereum known as also as PSC. So we can divide privacy at different in different dimensions. You have privacy of reads. I want to read the state the state of my state like privately. I want to I want to be able to also like transact privately with privacy of rights. And I want to be able to like be sovereign have sovereignty in terms of my own state and be able to prove over my state and therefore like have the the capability to do that like on my on my own like on my client.

So this is the main three track and there is like also a side track which is like network. How do we achieve network privacy? And therefore like having having the ability to transact fully privately. Um If you if you go to the maps or if you go to to our work like it's heavily like about privacy of rights and there is like a clear explanation about this is that as I introduced like we really like try to shape with institutional from institutional requirements and there is today a big push for tokenization and therefore we we we try to address this problem but I think that in the future we want to also have like a a full stack approach and try to map

solutions that are not just privacy of rights but how can you deploy from an institution? How can you deploy a yeah something that that also protect like end user privacy in terms of uh reads, for example. Um so, this is like uh some prototypes that we that we um So, we started to to prototype solutions like in in January. So, that's like some things that we we try to hack in the in the last uh quarter. Um again, I think the goal here for us is not to to make big breakthrough, but it's like uh more um educational. Uh so, um if there is an engineer from one of these big institutions that is like uh actually working on some project and wants to have like um a sense of what's

practical today uh to use on Ethereum, then he can come like consult like this uh proof of concept. We try to like be uh to use as much pattern as possible without being like too much opinionated. So, like for example, we we we use TE to explain like where does it where does it solve problems, but where it brings also like a lot of issues. Um like uh shielding, of course, like it's a is a is one of the main primitive. So, we we tend to develop a lot of use cases around that uh because it's we think it is very practical. So, we want to illustrate it uh the best as we can. Um and also we try to um to also address some use cases with uh with Validiums and how you can have like your own private um

private deployment. Um Yeah, so the goal of this in this presentation I'm I'm going to zoom in a little bit on shielding because um a lot for a lot of these institutions, they they're overwhelmed by about like the whole privacy space. Um and sometimes they come to us and they say they they want like something that is practical that they can um deliver tomorrow and they're they're completely lost. Like they they're like, "Yeah, I I've heard like that FHE is is actually the best solution or uh how can we leverage MPC or what's the specific threat model of of having a specific roll-up architecture?" So, um we we tend to develop a lot on shielded pool because it's just that from our own observation, it seems that it's very practical um and

that actually like that in in the short term like solve their problem pretty well. Um so, I'm I'm not going to enter too much into details, but it's just that uh just a rough specifications about like what what what a solution like this look like is just basically like uh the same old commitments and and unifiers. Um and well, the this whole construction like goes back to to the zero cash paper. Um and we just we just want to do like um more of this and more education toward the toward this uh because we think that it's uh again, it's uh it's practical and it's uh it's uh it's like the space also proved that uh a lot

of a lot of uh the privacy solution relies on it. So, we just want to like make sure that we explain it from the from the first principles. Another uh good point about like why shielding is interesting as a as a primitive to the to develop and to the build on is that it gives you like some um some compliance features that are interesting. Uh so, you have like a separation of concern. Uh you can have like a keys that are only meant for spending and keys that are like meant for viewing. So, um like if I'm an institution, if I'm a bank, for example, I can uh just share my viewing keys to uh the regulator and he can like audit my whole my whole um history of of transactions. Um and also you can use it as a as a way to

to carry uh to to have a private KYC. So, I can I can show that I'm I'm part I can generate a membership proof uh that I'm part of like the the the people that have been KYC'd and therefore like uh not sure not uh giving my my full identity there. Um Yeah so uh we we've applied this like to to uh to for example, a use case which is a private bond. Why we choose private bond is just because it's the in in this um movement of like tokenizing tokenize uh tokenize uh a lot of real world assets like bond is like the most simplistic form of of asset uh because it's very simple.

There is no not a lot of operation and just like you can uh for example, apply shielding to a bond and just like uh change the the the data structure about your commitment uh and uh and to make to to shoot like the the requirements of of a bond like in that case it's just for example, adding a a maturity date um and use the same the same shielding techniques uh on real world asset. Um and so, also one good point about like um um about shielding applied to institution is that usually a problem with shielding is that you you you need to farm this anonymity set. Otherwise, like uh you don't have like plausible deniability.

Uh but working with these institutions that want to deploy at a large scale is actually make this mission uh easier because uh they can farm their own anonymity set pretty quickly. Um Yeah, so basically you just you you you need like you need um uh this uh this component. You need you need a commitment scheme. You need as well as a an unifier scheme that they're all hash-based. Merkle Merkle membership proofs um and and then like the dual key architecture to have like both like spending capabilities, but as well as vendor vendor uh viewing capabilities.

Uh yeah, I think we want to we want to we want to show that we want to show that shielding is is could be seen as a as an abstraction um and just do more of of it like to for example, um help like if there is any initiative to like standardize the space, um have like open specifications about like how to handle this type and how to build this type of protocols. Um that's that's something that come that comes often like when discussing with with vendors is uh and vendors that are working with shielding is uh all right, like we should we should sit at a table and like um agree on like what what are the compliance rules for example, when deploying like shielded pools.

Um there's also like a research to be to be done on this on this primitive. Um of course, like as you have like to store like all these commitments and unifiers, like you you have a state load. So, there there exist um solutions to that, uh but still it's not like there doesn't seem that there is a golden standard about how to how to solve it. Uh post-quantum is also like a a topic. So, while all the primitives um of the abstract are post- quantum because it's mostly like um hashing, uh you still need to uh have like a ZK proving capabilities and therefore like there are there are some like benchmarking to do like in terms of what's the best prover like for the next uh quantum quantum era.

Uh censorship resistance. Uh today most of the shielded pools like rely on a on a on a relayer uh to to keep the plausible deniability uh property uh because you need you need to start on a fresh address and therefore you need someone to uh to pay the gas for you. Um there is a discussion about this new EIP 8141 with a frame transactions uh plus fossil maybe like the solution uh for that, uh but still it's uh it seems that the frame the frame EIP is not is not decided yet. Um and also like network privacy. Uh even if like your transaction is opaque uh in terms of uh of data, still someone can like see that all the all whole set of transaction coming from the

the same IP address, for example. Um So, on in terms of next steps, uh I think we want to double down on this framework with differentiating privacy for institution, but privacy from institution as well. Uh so, we want to make sure that uh the end user is not is not forgotten in the in the in the loop and make sure that when we discuss with with institution and doing education, we also like put awareness about like um uh the end user. So, for example, it could be like yeah, they they want to deploy their their private private roll-up and we we need to develop more technology to make sure that uh um uh user can force exit, for example, uh things like that.

Um so, yeah, what's next? Like more work on the end end-user privacy. Uh we also want like to um as I introduced like have like a more world stack approach like to think about like how institution needs to deal with identity and credential both internally, but also like on a on a more like public um uh public standpoint. Um governance and exit guarantees of of this of this stack deployed by institutions and as well as um how can what can be done like at the at the network level in terms of privacy. Uh and yeah, that's the that's the end of my presentation.

Uh again, like if if you're working like toward like more like privacy for for end user with like with like to connect, but as well as if you're solving like institutional privacy problems, I would love to have you like uh in the map and we're also like publishing all our uh work uh on our uh website. Um so, if you want to if you want to uh check this out, uh please uh um yeah, come in come in read our work. Thank you for your Uh thank you for your presentation. We almost run out of time, but maybe we can take one question.

Does anybody have any questions? Yeah. Thanks for the talk. At the beginning of your talk, you said your job is mainly about education for institution. Do you also uh help institutions implement a solution or do you leave that for the private sector like with Aztec and ZK ZK sync deploying Validium or do you also help uh companies and institution deploy uh private solutions? Yeah, I think the goal here is uh when if they if they are like uh really eager to put something into production or like do a more serious then it's why we are listing vendors and we are probably like just tell them, "Hey, we we are listing vendors like here are like people like who can you can follow up with and actually build

with." We are not positioning ourselves like in a in the the person that are going to help them deploy. We educate, we try to like uh like our mission is at the end of the day they need to like make their trade-offs with like the full picture about privacy and Ethereum and then they can like choose whatever vendor they want to like work with. Yeah. Thank you. Thank you.