## Private Transactions Research Team - January/June 2026

### Introduction

In 2026, the private transfers research team is expanding its mandate and growing its team to accelerate the delivery of new private transaction protocols on Ethereum. During the first six months of 2026, our five main goals will be to:

1. Help teams building on top of `plasmablind` ship to mainnet.
2. Deploy `sonobe` to an audited, production-ready version of our library. Intmax is developing zERC20s and we will be supporting them in the endeavour.
3. Propose a new wormhole design, probably based off beacon chain deposits.
4. Research and if possible implement one-time or pay-per-use programs using garbled circuits and extractable witness encryption. Such constructions are very attractive and recent research seems to be converging to practical instantiations making it possible to run decryption protocols conditional on some onchain state.
5. Build expertise on remotely familiar cryptography or hardware based primitives and contribute back to some well-known open-source libraries.

While Chengru will help teams like Intmax leverage `sonobe` to productionalize their privacy protocols, Mohammad (who is joining the team in february) will work on how to leverage existing (or possible) L1 primitives to bake privacy within Ethereum itself. We are also meeting together in Tokyo in march, with several others researchers and industry teams (Intmax) to further our collab on `sonobe` and ideation around this research vertical.

### Topics

We list here the five topics that we will be working on from January to end of June 2026.

#### 1. Plasmablind

Researchers: Chengru, Pierre

We aren't sure of what PlasmaBlind reception is going to look like. When decentralized, PlasmaBlind is able to accomodate around 300 to 500 TPS, with instant proving times on low-end devices. This TPS can be much higher, the setup's bottleneck being mostly the aggregator's available network's bandwidth. On this project, our agenda might depend a lot on whether the scheme gets some teams getting interested into implementing it in the months following the work. We are planning for:

1. Finishing paper writeup
2. Conference submission
3. Potential support for external teams

#### 2. Sonobe, Intmax and zERC-20

Researchers: Chengru, Pierre, Yutaka Hashimoto (Intmax)

Intmax has started to investigate and integrate folding schemes into its proving stack. They are asking for guidance on auditing and integrating `sonobe` for their zERC20 implementation. To make Sonobe credible for production use, we want to do a security hardening phase. Our role will be to drive the preparation work with the Intmax team, support auditors and land resulting changes. To reduce the audit's surface area, we will narrow (or even sunset) most of the supported frontend targets of `sonobe` and only focus on the few that matter. The intent is to make the core correct, auditable, and easy to operate. Once done, we will ship a set of improvements to `sonobe`: adding lookup support, improving devex, tightening performance on some hotpaths. We are planning for:

1. Shipping to `main` the current `dev` branch
2. Sonobe AI and human assisted audit
3. Intmax team support/advisory
4. Upgrades: lookup support, optimizations, devex improvements

#### 3. New Wormholes

Researchers: Mohammad, Pierre

While mixers are ones of the most popular tool individual users leverage to get privacy on Ethereum, the state of the ecosystem is disappointing: we still leverage CEXes to initialize fresh wallets, Tornado Cash remains sanctioned with detectable interactions, most protocols don't feature plausible deniability. Wormholes offer an attractive route to enshrining privacy at the base layer. However, we discovered our previous wormhole approach is broken and [wrote a blog post](https://ethresear.ch/t/wormholes-and-the-cost-of-plausible-deniability/23728) documenting the issue. Over the next six months, we will treat wormholes as a redesign effort. We want to re-derive security goals, identify where the earlier construction failed, and propose a new path for wormhole L1 enshrinement.

One avenue for work is to design a wormhole leveraging beacon deposits. We would also like to propose an implementation.

1. Research phase for leveraging beacon chain deposits
2. ETHresearch post and gathering feedbacks
3. Potential EIP and implementation

#### 4. One-time programs and stealth mixers

Researchers: Chengru, Mohammad, Pierre

Design and prototype mixers using one-time programs (OTP). This research vertical is particularly attractive since it could help us get non-custodial, plausibly deniable mixers. One-time programs can be built from the combination of garbled circuits and extractable witness encryption. While the practicality of witness encryption for general NP problems remains challenging, recent research points out that combining signature-based witness encryption with extractable witness encryption for KZG commitments can lead to something called T+1 extractable witness encryption for blockchains, an almost practical primitive making the decryption of a secret contingent on some chain state. However, three questions have been left open: (1) reduce or remove the signature size blowup, (2) make it possible to run the program at any time (T+N), (3) have a pq version.

1. Octopus contracts, signature-based witness encryption and garbled circuits research ([2025/1064](https://eprint.iacr.org/2025/1064.pdf), [2025/443](https://eprint.iacr.org/2025/443.pdf)).
2. Mixer design, scheme improvements based off trying to answer questions (1) and (2)
3. ETHresearch post and gathering feedbacks

#### 5. MISC

Researchers: Chengru, Mohammad, Pierre

Some other miscellaneous work that we are keeping an eye on:

1. TEEs have attractive affordances for privacy transfers, but we don’t yet have the depth needed to be confident in “TEE-assisted” protocol design. In 2026 we want to close that gap: read and summarize the key literature, map the known failure modes and adversary models, and produce an internal (and possibly public) write-up that states what we believe TEEs are good for, what we will not rely on them for, and what protocol patterns are actually defensible.
2. Various cryptographic primitives implementation: proxy re-encryption, helping ship `spongefish` to have R1CS support, proof aggregation protocols for different privacy-protocol to share a single onchain proof (gas costs reduction).
3. We are also more generally trying to keep some buffer for applications building.

#### Addendum: strengthening our relations with academia

We’ll strive to continue our existing relationships to be sustained, high-leverage collaboration. When appropirate, we will also sollicit the academic secretariat's help.

- **EPFL CompSec lab.** We already have good momentum (e.g. hash-based scheme directions). In 2026 we would like to convert this into one or two concrete joint outcomes: a co-authored write-up/paper or another shared implementation. I'm trying to seduce them to get interested about privacy protocols.
- **Srinath Setty and the folding/IVC community.** We are continuing our ties with Srinath by doing regular syncs, sharing our implementation learnings early and collaborating on open problems where our constraints can inform his research priorities.
- **Arkworks** Contributing back all the work which has been done so far on arkworks (spongefish, folding schemes, nary trees, new hash functions..)