DSS 2026 / Day 2 / Amber Hall
Are MPC wallets ready for the post quantum age?
Speaker credit under review
The published video credit and the programme differ. This recording is not assigned to a programme speaker while the credit is under review.
Source: YouTube automatic caption track · Exported 2026-11-16. Timestamps link to the source video. Cue end times were not supplied.
Transcript
0:00Per so good afternoon everyone. I'm Yun
0:02from Fortify and today we'll be talking
0:04about whether your MPC wallets are
0:06actually ready for the postcoh. So to
0:09starters we are fortify by paxos. We do
0:12NPC wallets that for in institutional
0:16usages where NPC is at our core that
0:19then keys are split and never
0:21reconstructed to sign. Uh disclaimer
0:23this talk was meant to be a a joint talk
0:26with my CEO Dimma but he wasn't able to
0:28attend. So today we will discuss why the
0:31heck should we care about quantum
0:32computing? What's the actual quantum
0:34thread and what are the new signatures
0:36that we're considering? what the
0:38blockchain are planning to do and how
0:40does postquantum look in the NPC world
0:42and well if it's not ready what can we
0:44actually do cool so why the heck should
0:47I care now to be honest it's quite
0:49simple quantum attacks are getting quite
0:51closer now like large companies such as
0:54Google are targeting 2029 to migrate to
0:56fully postquantum schemes and ECDA is
0:59going to be dupregated by 2023 by 230
1:02sorry so it's quite around us around the
1:06corner to be honest And moreover,
1:08research and standardization of new
1:10schemes actually take time, especially
1:12the MPC schemes that really comely come
1:14after only the signing schemes. And
1:17moreover, the chain migrations also take
1:19time. So, we need to start early. But
1:21what actually happens once we have a
1:23quantum computer? First and foremost,
1:25almost all public key cryptography is
1:27going to be broken. Using Shor's
1:29algorithm, we given a public key, you
1:31could simply take a look at it and say,
1:33well, I can break that and find the
1:34private key quite instantly. So that
1:37means that all our ECDSA, EDSA and other
1:40elliptic curves or even RSA is going to
1:42be broken. And on the other hand,
1:44there's also the symmetric keys such as
1:46SHA and AES. They're going to be
1:48weakened,
1:49which means that we need larger keys,
1:51but we can still use them. So the major
1:54blockchains are using skins that are
1:56going to be broken. Then we need to
1:57replace them. But what are we going to
1:59replace them with? Well, there are many
2:01options er such as Magnus actually said
2:05there are many schemes such as hashbased
2:07signature which we can use specifically
2:09SLHDSA and there are the latest ones
2:11such as MLDDSA and Falcon which these
2:14schemes are the most
2:16used schemes to be honest the most
2:18looked at because they are the N
2:19standards there are other schemes which
2:21we won't discuss and there are other
2:23schemes which were really leading in the
2:26candidates of the N standardization and
2:27they were simply broking along the Okay,
2:30so you can think of it as it's quite
2:32hard to actually think of and invent new
2:34schemes and new cryptographic
2:35mathematics. So I won't discuss the
2:38hashbased signature a lot again thanks
2:40to Magnus which helped me a lot here
2:41this time but in very short terms you
2:44have a hash value you have a secret key
2:46you simply hash it a number of times say
2:48hey this is my signature and then you
2:50give it to some verifier which hashes
2:51the remaining number of times compare it
2:54to your public and says well you're okay
2:55that signature is fine on top of that
2:58you build more sophisticated scheme such
3:00as SLHDA which I I like to think of it
3:03as a tree signing the root of other
3:05three. Now a very important point here
3:08is that it's considered a conservative
3:10option. Why? Well, we are relying on
3:13hash hash schemes which we know and love
3:15and use every day. And moreover, as you
3:18can notice, I never said which hash
3:19function we actually are using. So if
3:21the hash is broken, simply change the
3:23hash function. But you don't need to
3:24change the scheme. Now it's not all
3:27perfect. It has its own weaknesses such
3:29as being slow and also it has large
3:32signatures. like it's almost 100 times
3:34larger than SDSA.
3:37On the other hand, we have the latises
3:39which is the new cool method on the
3:41field of cryptography. What's lises?
3:43Very complicated. In short bases, it's
3:46simply think of it as linear algebra.
3:48You have your basis vector. It's long
3:50and skewed and angly. And then your goal
3:52is to find a very small vector that
3:54simply is hard to find giving those
3:56bases. In the picture here, we see that
3:58it's 2D. Actually, it's a thousand
4:00dimension. So, it's quite hard. And
4:02there are the most two interested
4:05schemes which are MLDDSA and Falcon.
4:07LDSA relies on noisy linear equations
4:09and Falcon relies on structured latises.
4:12Where are these? Well, that's not
4:13important for our talk today. And
4:15there's a very interesting here point
4:17here which is these math are actually
4:19quite debated. Like if you ask the
4:21German federal office, they will say
4:23well it's nice, it's cool, but please
4:25use it in a hybrid form. Please use some
4:27kind of ECD say currently because we are
4:29unsure that it's actually safe. or even
4:31there are people in the industry that
4:33simply say well please avoid structured
4:35mathematical assumptions entirely. I
4:37think it says something a bit or two
4:39about these schemes and also it has its
4:42own weaknesses such as being large in
4:44public keys but it also has new
4:46strengths which is being fast and being
4:48small in signature size. Cool. Now let's
4:51take a step back and look at the point
4:52of view of a blockchain. Well, what's
4:54important for them? There are three main
4:56points. First is the hardness which I
4:58find the most interesting as a
4:59cryptographer. And there is also
5:01signature size which and key size also
5:03which is quite important for them
5:05considering that we want small
5:06signatures and keys. And there's also
5:08the aggregation points. So what each
5:11blockchain is considering? Well, it's
5:13quite diverse to be honest. They are the
5:15ones that are considering hashbased
5:16because of its conservativeness such as
5:18arc essentially which I really like
5:20because you can now use pre-ompiles of
5:22SLHDSA specifically and their mainet
5:25which is quite nice and there's bitcoins
5:27which in simply are considering a forked
5:30of SLHDSA
5:31similar to sphinx minus essentially
5:35which reduces the size signature sizes
5:38and also make it faster. On the other
5:40hand, there are ones which are
5:41considering the fingering, sorry,
5:44favoring the size such as Salana and
5:46Tron and Ethereum. As Magnus again said,
5:49they are quite generic. They're saying,
5:51well, use what the heck you ever you
5:52want. You can use Falcon, MLDDSA, ECDA,
5:55sorry, SLHDSA. Eventually, currently,
5:58you can actually use it in Ethereum, but
6:00eventually it will be native with
6:02pre-ompiles with EIP8141.
6:04And for the consensus, they are
6:06considering mostly the hashbased again
6:08for its conservativeness.
6:10Now let's even take a step back. What
6:12are what is going to happen with NPC
6:14wallets? Well, a bit about NPC wallets.
6:16For those who don't know, NPC is simply
6:18the idea of having multiple parties
6:20perform a computation where your inputs
6:23are simply private. Think of it as a
6:26multiple parties which know which want
6:27to know who is richer to be honest like
6:29it can be any math problems that you
6:31want to solve. And in specifically in
6:33the signing world of crypto, it's the
6:36idea of having number of parties which
6:37each hold a share of the key of the
6:40private key and eventually they need to
6:42communicate together perform some kind
6:43of protocol computation eventually to
6:46create a new standardized a new standard
6:48signature which is invincible on chain
6:50compared to some something like multisig
6:52which you know how much people actually
6:54were used to sign that signature. And
6:56the primary bottleneck here is actually
6:58the communication. The math here, how
7:00much it takes to create a signature or
7:03to do the computation relies heavily on
7:05how many rounds do I need to send to the
7:07other people. How many packets do I need
7:09to send them? Because you're doing a
7:10joint computation. You don't simply send
7:12a message and then boom, the secondary
7:14is ready. So there are three main groups
7:17of computation of steps essentially.
7:19There are the linear steps which are
7:20simply local operations that you do.
7:23They're quite cheap. They're local
7:24operations. you don't need to send
7:25anything to the network. On the other
7:27hand, there are some things like multip
7:29multiplying secrets. These require
7:31simply a round or two based on the
7:32scheme of MPC that you're using. And
7:34then there's the interesting ones, the
7:36nonlinear operations such as hashings
7:38which require a lot of communication
7:40rounds to do. Essentially, you would
7:42need a lot of communication rounds
7:43sending to each other waiting for the
7:45response until to be able to perform
7:47that signing. And the important point
7:49here is today's signatures they are
7:51using mostly linear operations. So they
7:53are quite interactive to be honest. But
7:55the postquantum schemes well there are a
7:57different story. They are more nonlinear
8:00rounds and they require a lot of more
8:03nonlinear operations to do so. So you
8:06can assume that's going to be quite
8:08slow. How slow? Well, that's exactly
8:10what we wanted to see. And we decided to
8:13take a look of what would happen if now
8:15there was a quantum computer. What how
8:17how would we actually do and perform
8:20with these? So we took the most
8:22important known protocols in the MPC
8:24world considering the postquantum war
8:27and benchmarked them benchmarked them on
8:29the two-party signing. We considered two
8:31different deployment models. First the
8:33same region which we have two servers
8:35sitting next to each other which high
8:37with high bandwidth network and on the
8:39other hand you have the phone and cloud
8:40similar to how we do at Fortify which we
8:42have an enclave and a mobile which
8:43communicate and here your bandwidth is
8:46quite more constrained and we time two
8:49things. First the offline the offline
8:51phase which is simply preparation of
8:53your message that you're going to sign.
8:56You can do this before you need you know
8:57what message you're going to sign which
8:59in case of blockchain what is your
9:01transaction that you're going to sign
9:02and the only phase that once you have
9:04your transaction that you want to sign
9:05how much time would you need until you
9:07sign it. So now we're going to go over
9:10the most important protocols and see how
9:13it actually deployed. So starting with
9:14MLDDSA there is the myth world protocol.
9:17Quite a nice one. And the problem with
9:19MLDDSA is the idea that even once you
9:22have the signature of your transaction,
9:24you can't simply broadcast it away
9:26immediately. You need to check that it's
9:29it has some kind of requirements that
9:30doesn't leak your information. And the
9:33idea here is that in the NPC state, in
9:36the NPC regime, you need to actually
9:37verify that it's okay to rebroadcast,
9:40but you have other people that you need
9:42to also trust. So you can't trust them.
9:44So what did they do? They essentially
9:45took a different approach was simply
9:48each party would consider his own part
9:50of the signature and say okay it
9:52verifies that it's small enough let's
9:54say and then broadcastes to the other
9:56people to simply build up the signature
9:59which enables it actually to be quite
10:01interactive on each and every different
10:03network and eventually it's quite
10:05practical up to the six parties but you
10:08know most of the deployment modes is
10:09actually enough for and on the other
10:12hand there is threshold falcon threshold
10:14falcon is quite write new protocol and
10:16it's a breakthrough to be honest for
10:19those who know about about Falcon. Well,
10:21Fal Falcon relies on floating points.
10:23Floating points and NBC are they aren't
10:26friends. They hate each other. They
10:27don't work together at all. So, what did
10:29they do? They essentially targeted a
10:31different random sampler. Instead of
10:34Gaussian, they used Klein. What does
10:35that mean? Not that important, but it's
10:37simply that they use integers and they
10:40also target batch signs. The number that
10:42we are seeing here are essentially when
10:43you want to sign a single message and
10:45you can see that it's not quite so
10:47interactive such as MLDDSA but still
10:50quite usable on the server server side
10:52on the mobile and server well not so
10:55much but if you consider bat signing
10:58like let's say I don't know signing 200
11:00signatures at once it can be actually
11:02quite usable mostly on the server server
11:04side and there's a small caveat which
11:07you need a trusted dealer like some kind
11:09of third party which comes and splits
11:10your keys and shares and does all the
11:12first pre-processing state. And now the
11:15fun part begins for me which is SLHDSA.
11:19Up until last two months, there weren't
11:22any MPC protocols at all for SLHDSA and
11:26there was actually a paper which
11:28released which used fully homorphic for
11:31for those who know which required eight
11:33GPUs such as RTX 490 and could sign a
11:37message in 3 minutes. Well, nice, cool,
11:40but not so interactive. And we have been
11:42doing doing our own research to be
11:45honest and as a HD and we were able to
11:47sign in two party mode between a server
11:49and a server or a mobile and a server in
11:51less than a second. And how did we do
11:55that? Well, we're going to publish a
11:57paper later on in a couple of few weeks.
11:59So stay tuned for that. But in short,
12:02some of the key ideas that we did is the
12:04fact that SLDA can be quite paralyzable.
12:07So what happens if you paralyze it with
12:09GPUs and quite heavy optimizations?
12:13Well, you can get some interesting
12:14things. What are the caveats from our
12:16side? Well, we're mostly targeting the
12:18two-party signing although it can be
12:20done also for multiple party and we're
12:22targeting the shape parameter sets for
12:24those who know that hash function and
12:27specifically because it's quite MPC
12:29friendly.
12:31Now let's say that today a a quantum
12:34computer is deployed and a lot of the
12:36other schemes other than the ones that
12:38we log they don't have an APC protocol.
12:40Well can what can you do actually
12:41instead there are many options such as
12:43onchain multiig but then you would make
12:46it visible how many signage you need and
12:48how many verifiers you need or secret
12:50sharing. What if you simply take your
12:52private key and secret share between
12:53number of parties and simply have some
12:55kind of trust in environment signing
12:57that it's okay but you know if your
12:59trusted environment is compromised then
13:00your key is compromised not as NPC which
13:03you need to compromise all the parties
13:05at once there are also snack
13:06aggregations nice but they're quite
13:08heavy and require specific operations on
13:11chains and there are hybrids which I
13:13really like I even like it so much that
13:16we're going to look at a case study what
13:18happens if we actually need to use
13:19hybrids we can do something quite
13:21interesting. You can use the MPC for the
13:23classical side for the ACDSA side and
13:26then you can also use secret sharing
13:28specifically to pro like to secret share
13:31your postquantum key and then use it on
13:33a trusted environment. Then you would
13:35actually need like a way to break let's
13:38you still have the NPC because you need
13:40the classical part to sign with to sign
13:42the classical part like you need the NPC
13:44on the other side for the postquantum
13:46you need the two secret shares. Now
13:48there's the point that it's
13:49reconstructed on the trusted
13:50environment. So if the trusted
13:52environment is compromised, you still
13:54need the hybrid signing unless a quantum
13:57machine actually compromises you, then
13:59you're simply screwed and you can't do
14:01anything. So these are the points that
14:03if we actually consider
14:06more schemes essentially require more
14:08NPC research and the alternatives are
14:11not exactly a replacement. Each one has
14:13its own problems and chains should
14:15actually weigh NPC when they choosing
14:18the scheme because when they're choosing
14:19the schemes they are quite determined
14:22how much the MPC wallet can actually
14:24follow in the postquantum world. Yeah.
14:26Thank you.
Source and accuracy details
- Names, organizations, technical terms, code and numbers may be misrecognized.
- Source wording and disfluencies are preserved; no editorial corrections were applied.
- Export contains whole-second cue starts only; cue ends and subsecond timing are unavailable.
- A full caption-track export does not establish that every spoken word is represented.
- Automatic YouTube captions; not human-verified against the audio.
Source export SHA256: 54063e7345327fdb7383b24447ffe5503744dc387fa6079dfff1577cbb7189db