Ethereum Institutional · Research

Privacy

Confidential tokens are the center of privacy on Ethereum, and privacy here is thriving in many forms, on Ethereum mainnet, on Layer 2s and on private networks. The map shows what each design hides from the public; Focus on picks out the designs that fit what you need, and the institutional wish list shows which come closest.

Live on an EVM ledger or a Layer 2 that settles on Ethereum mainnet On a public testnet
Regulatory lensesundefined
Inside the definition (thick line) Depends, exempt or restricted (long dashes) Outside the definition (faded)

Privacy on a shared ledger

Hover over a circle or a design to read what it hides; click to pin it here with all its facts. Use Focus on to pick what matters to you, or start from the institutional wish list.

What institutions are asking for

Confidential tokens are what institutions want from privacy, with requirements that differ from an individual’s. No design meets them all yet: pick the institutional wish list in Focus on to see which come closest.

Permissioned, with a notary

In our work with institutions, permissionless privacy, open to anyone with no check on who enters, is treated as a reputational risk, and the concern travels with the service: a pool, network or named service that also serves permissionless privacy raises it even when your own deployment is permissioned, while software you deploy for yourself does not.

More

Institutions want privacy among verified participants holding verified assets, with a notary (the institution, its regulator, or law enforcement on an order) who can decrypt in a way that leaves an on-chain notice: the auditability Project Epic names beside anonymity and confidentiality. Standard setters ask firms to manage anonymity-enhancing technology or stay away from it; US Treasury, for its part, told Congress in March 2026 that lawful users may use mixers for financial privacy.

See also Identity map →

Sources (3)

undefined; undefined; undefined

Permission comes from identity

Permissioned privacy needs a check on who enters, and that check is identity: know-your-customer, know-your-business or know-your-agent results carried onto the ledger as an allowlist, an attestation or a credential proof. How it is carried matters here too: an allowlist is public even when the amounts are not, while a zero-knowledge credential proves the check passed and nothing more.

More

Identity needs privacy of its own: data-protection rules keep personal data off the ledger, so only a salted commitment or a proof that the check passed should reach it, which the EU data-protection board's guidelines allow for and which stays hidden against quantum attacks, because it rests on hash functions. The Identity map sets out each mechanism and what it reveals.

See also Identity map →

Sources (1)

EDPB, Guidelines 02/2025 on processing of personal data through blockchain technologies, version 2.0 (adopted 7 July 2026), paragraphs 52 to 54 (salted or keyed hashes and perfectly hiding commitments) and 104 (not advisable to record clear text, encrypted or hashed personal data, given erasure) (checked 2026-11-16)

Any EVM ledger, any token

One design for private, consortium and public ledgers under any consensus, carrying ERC-20 stablecoins and deposit tokens, permissioned securities (ERC-3643, ERC-7943), non-fungible tokens and vault shares, with the issuer’s freeze, forced transfer and recovery intact.

Open, verifiable and built to last

Open-source code on open standards, independently audited, with every private transfer verified by the chain rather than trusted to an operator’s signature, and nothing on the public record a quantum computer could decrypt later. Every component (prover, indexer, key service, relayer) is one you can run yourself, with no dependence on another network and no fee beyond gas, and it works with keys your custodian already manages. If the privacy layer fails, you should lose privacy, not funds.

Privacy across the whole workflow

A confidential token does not make the workflow private. Check every layer a transaction touches.

Getting in and out

Deposits into and withdrawals from a shielded pool are public, and the address that pays gas links to the one that acts unless a relayer or an ERC-4337 paymaster pays instead.

Sources (2)

Privacy Pools docs, deposit and withdrawal (public deposits; direct versus relayed withdrawals) (checked 2026-10-26); ERC-4337, Account Abstraction (paymasters pay gas on the sender’s behalf; Final) (checked 2026-10-26)

Data kept off chain

Keeping data off chain (a private group, a validium or a notary’s ledger) hides it from the public, but the operator and whoever holds the data still see everything, and you rely on them to keep it available.

Sources (1)

ethereum.org, Validium (off-chain data remains visible to the operator) (checked 2026-10-26)

Messages beside the ledger

Payment messages travel off chain beside any confidential token, and ISO 20022 calls for structured identification of every party.

Sources (1)

CPMI, Harmonised ISO 20022 data requirements (October 2023) (checked 2026-10-26)

The Ethereum roadmap for privacy

Privacy is a stated goal of the Ethereum roadmap, built in three pillars: private writes, private reads and private proving. Dates are targets, not guarantees; items move a reading on this page only once they are live.

Live today

Stealth addresses and their registry (ERC-5564 and ERC-6538, both Final); account abstraction (ERC-4337, and EIP-7702 since Pectra in May 2025), which lets a third party pay gas so a private address need not be funded from a linked one; and the confidential-token interface ERC-7945, Final on 7 October 2026. ERC-7984, the other confidential-token interface, is a Draft.

ethereum.org →

Sources (5)

ERC-5564, Stealth Addresses (Final, 28 May 2024) (checked 2026-10-26); ERC-6538, Stealth Meta-Address Registry (Final, 25 June 2024) (checked 2026-10-26); ERC-7945, Confidential Transactions Supported Token (Final, 7 October 2026) (checked 2026-10-26); ERC-7984, Confidential Fungible Token (Draft) (checked 2026-10-26); EIP-7702, Set Code for EOAs (Final; live since Pectra, 7 May 2025) (checked 2026-10-26)

Coming next: private transfers without a relayer

Frame transactions (EIP-8141, scheduled for the Hegotá upgrade) let a transaction’s authorization be any program, including a zero-knowledge proof. Two drafts make that fast and reliable for privacy protocols: keyed nonces (EIP-8250) let many users spend through one shared sender at once, each spend’s nonce derived from its nullifier, instead of queuing behind a single nonce; and recent roots (EIP-8272) let a spend prove against a privacy tree root from a recent slot, so nodes can admit it to the public mempool without a relayer.

More

All three are Drafts, running together on a public Hegotá privacy testnet (chain 8141, the ethrex client) since 11 August 2026, where a transaction carrying a proof costs about 352,800 gas.

Sources (5)

EIP-8081, Hardfork Meta: Hegotá (scheduled, considered and declined items) (checked 2026-10-26); EIP-8141, Frame Transaction (Draft) (checked 2026-10-26); EIP-8250, Keyed Nonces for Frame Transactions (Draft, 16 April 2026) (checked 2026-10-26); EIP-8272, Recent Roots for Frame Transactions (Draft, 15 May 2026) (checked 2026-10-26); Hegotá privacy testnet (ethrex, chain 8141): faucet and changelog (checked 2026-10-26)

Built into the protocol: not yet

A protocol-level shielded pool (EIP-8182) was declined for Hegotá; an enshrined encrypted mempool (EIP-8105), which would hide transactions until their order is fixed, is a Draft.

Sources (3)

EIP-8081, Hardfork Meta: Hegotá (scheduled, considered and declined items) (checked 2026-10-26); EIP-8105, Universal Enshrined Encrypted Mempool (Draft) (checked 2026-10-26); Ethereum Foundation, EF Protocol: current and emerging priorities (7 September 2026) (checked 2026-10-26)

Who is building it

The Ethereum Foundation’s Privacy Stewards of Ethereum (formerly PSE) lead the work, including the Kohaku wallet toolkit (alpha) and a public benchmark of private-transfer protocols.

PSE →

Sources (4)

PSE, Roadmap: 2025 and beyond (Privacy Stewards of Ethereum) (checked 2026-10-26); Ethereum Foundation, The EF’s commitment to privacy (8 October 2025) (checked 2026-10-26); ethereum/kohaku repository (alpha, unaudited) (checked 2026-10-26); Vitalik Buterin, A maximally simple L1 privacy roadmap (11 April 2025) (checked 2026-10-26)

Confidentiality that has to last

Ledger data is permanent, so ask how long each secret must stay secret, and what a quantum computer, or a new piece of mathematics, would expose.

What lasts and what does not

Fully homomorphic encryption is built on lattices, believed safe against quantum attack, and hash-based STARK proofs are too. Pairing-based SNARKs are not: a quantum computer could forge proofs, though a stored proof does not reveal its secret inputs. The notes many shielded pools encrypt on chain with elliptic-curve keys could be decrypted later: harvest now, decrypt later. A design that keeps only salted hashes or state roots on chain leaves nothing public to decrypt, though its signatures and the channels carrying the private data stay classical.

Quantum →

Sources (4)

Zama TFHE-rs, security and cryptography (LWE, quantum-resistant) (checked 2026-10-26); ethereum.org, Quantum resistance (SNARKs, KZG and STARKs) (checked 2026-10-26); Zcash ZIP 2005, quantum recoverability (balance versus privacy against a quantum adversary) (checked 2026-10-26); NIST IR 8547 initial public draft (harvest now, decrypt later; 2030 and 2035) (checked 2026-10-26)

Not only quantum: the mathematics is moving too

The assumptions behind privacy cryptography can weaken without any quantum computer. In July 2026 an AI model found an unexploited symmetry in the lattice behind HAWK, a third-round candidate in NIST's process for additional post-quantum signatures, and cut the demonstrated cost of recovering a HAWK-256 key from an estimated 2⁶⁴ operations to about 2³⁸; the attack stays exponential and is specific to HAWK, nothing deployed had to change, and the HAWK team withdrew the scheme the next day.

More

The same work made a known attack on a seven-round, reduced AES 200 to 800 times faster, still impractical. Lattices are what fully homomorphic encryption and most post-quantum signatures rest on, and Ethereum's own researchers said in October 2026 that the concrete security of lattices should be expected to take hits from AI-driven mathematics in the next two years, with hash-based constructions preferred where they fit and larger lattice parameters where they do not. Ethereum Institutional reads it as the Quantum pages do: prefer the fewest assumptions, keep parameters conservative and schemes replaceable, and assume that whatever sits on a public ledger will be read by better mathematics than exists today.

Quantum, the algorithms and the AI threat →

Sources (3)

Anthropic, Discovering cryptographic weaknesses with Claude (28 July 2026): an unexploited automorphism in HAWK's lattice, key recovery for HAWK-256 demonstrated at about 2³⁸ operations against an estimated 2⁶⁴, the attack still exponential and specific to HAWK; a 200 to 800 times faster attack on seven-round AES-128, still impractical; no production software has to change (checked 2026-11-10); NIST, Nine candidates advance to the third round of additional digital signatures (14 May 2026, updated 29 July 2026: the HAWK team withdrew its algorithm on 29 July 2026 after the AI-found vulnerability) (checked 2026-11-10); Decrypt, Vitalik Buterin on AI-accelerated mathematics and lattices (8 October 2026; secondary): the concrete security of lattices expected to take serious hits from the next two years of AI math; hash-based preferred; larger lattice parameters; no rushed wallet moves (checked 2026-11-10)

What regulators have said

Dates and holdings only. Informational, not legal advice.

Standard setters and the EU

The FATF’s updated guidance of 28 October 2021 says virtual asset service providers must be able to manage the risks of anonymity-enhancing technologies, including mixers and privacy wallets, or should not engage with them. The EU anti-money-laundering regulation bars anonymous accounts and accounts allowing anonymity-enhancing coins from 10 July 2027.

Sources (2)

FATF, Updated Guidance for a Risk-Based Approach to Virtual Assets and VASPs (28 October 2021), paragraph 174 (checked 2026-10-26); Regulation (EU) 2024/1624 of 31 May 2024 (OJ L, 2024/1624, 19.6.2024), Article 79(1): no anonymous crypto-asset accounts or accounts allowing anonymisation, including through anonymity-enhancing coins; Article 90: applies from 10 July 2027 (checked 2026-11-09)

United States

OFAC designated Tornado Cash on 8 August 2022. On 26 November 2024 the Fifth Circuit held that its immutable smart contracts are not property under IEEPA, and Treasury removed the designation on 21 March 2025.

More

Treasury’s March 2026 report to Congress under the GENIUS Act says lawful users may use mixers for financial privacy on public blockchains. The criminal case is separate: in United States v. Storm (S.D.N.Y.) a jury convicted a Tornado Cash co-founder on 6 August 2025 of conspiracy to operate an unlicensed money transmitting business and did not reach a verdict on the money laundering and sanctions conspiracy counts; a retrial on those counts is set for 26 April 2027, with a motion for acquittal pending.

Sources (6)

US Treasury, Tornado Cash designation (8 August 2022) (checked 2026-10-26); Fifth Circuit, Van Loon v. Department of the Treasury, No. 23-50669 (26 November 2024) (checked 2026-10-26); US Treasury, Tornado Cash delisting (21 March 2025) (checked 2026-10-26); US Treasury, GENIUS Act report to Congress on innovative methods (March 2026) (checked 2026-10-26); US Attorney's Office for the Southern District of New York, press release of 6 August 2025 (republished by IRS Criminal Investigation): Tornado Cash co-founder convicted of conspiracy to operate an unlicensed money transmitting business (checked 2026-11-09); Decrypt, Tornado Cash retrial moved to 26 April 2027 (26 August 2026; secondary): the deadlocked money laundering and sanctions conspiracy counts; motion for acquittal pending (checked 2026-11-09)

Central banks and supervisors

The BIS Project Agorá prototype (report of 27 May 2026) used Hyperledger Paladin for token-level and workflow-level privacy. The HKMA’s e-HKD pilot tested Zeto and Anonymous Zether on public chains (October 2025). The FCA and the Bank of England wrote in May 2026 that “technological advances mean that public chains can now also maintain privacy standards”.

Sources (3)

BIS, Project Agorá report (27 May 2026; Paladin Noto and Pente) (checked 2026-10-26); HKMA, e-HKD Pilot Programme Phase 2 report (October 2025) (checked 2026-10-26); FCA and Bank of England, The future of tokenisation, call for input (May 2026), paragraph 3.18 (checked 2026-10-26)

Help shape this map. Open an issue to suggest a design, a correction or a deployment the map has missed. The maps are edited in Ethereum Institutional's platform repository and mirrored here, so an issue reaches the editors faster than a pull request against the mirror.

This map is an Ethereum Institutional house view for discussion, not legal advice. © 2026 Ethereum Institutional. Licensed under CC BY-SA 4.0 (share and adapt with credit, under the same license). Version 1137e34 · built 2026-11-16