Ethereum launches Stealth Addresses group with ML-KEM
The Ethereum Foundation launches the Stealth Addresses working group with Umbra, Fluidkey, and Trezor to standardize hybrid ML-KEM post-quantum privacy.

Transactional privacy on Ethereum has just taken a decisive architectural leap toward long-term quantum resilience. The Ethereum Foundation has officially announced the formation of the Stealth Addresses working group in direct collaboration with the engineering teams behind Umbra, Fluidkey, and hardware wallet manufacturer Trezor. The core mission of this technical alliance is to evolve the current ERC-5564 standard into a unified specification incorporating ML-KEM (FIPS 203) post-quantum key encapsulation mechanisms and native hardware custody integration across the Web3 ecosystem.
Within the transparent account model of the Ethereum Virtual Machine (EVM), reusing a single public 0x... address to receive payroll, donations, institutional settlements, or commercial payments exposes a user's entire net worth and financial relationship graph to any external observer. Stealth addresses solve this exposure by allowing a sender to non-interactively generate a unique, single-use destination address for every transfer, ensuring that only the legitimate recipient can detect the incoming payment and compute the private key required to spend those funds, while preventing blockchain analytics firms from linking the fresh account to the receiver's primary identity.
The quantum paradox of announcement registries on public blockchains
Despite the milestone achieved by ratifying ERC-5564 (Stealth Addresses) and ERC-6538 (Stealth Meta-Address Registry), cryptographers at the Ethereum Foundation, Umbra, and Fluidkey identified a structural vulnerability that uniquely affects privacy on immutable public ledgers. In traditional network communications, an adversary must actively tap communication lines in real time to execute a passive harvesting attack; on a public blockchain like Ethereum, however, every event emitted by the registry contract (Announcement) is replicated and stored publicly forever across thousands of archival nodes worldwide.
Under the classical ERC-5564 design over the secp256k1 elliptic curve, a recipient publishes a meta-address consisting of a spending public key ($P_{\text{spend}}$) and a viewing public key ($P_{\text{view}}$). When a sender transfers assets, their wallet generates an ephemeral key pair ($r, R = r \cdot G$), publishes the ephemeral public key $R$ inside the announcement contract logs, and computes the shared secret $S = r \cdot P_{\text{view}}$. From $S$, it derives the one-time stealth destination address:
$$P_{\text{stealth}} = P_{\text{spend}} + \text{keccak256}(S) \cdot G$$
The vulnerability lies in privacy durability. Even if a user spends or migrates their funds to a quantum-resistant vault the day quantum processors threaten spending signatures, the historical on-chain record of the Announcement event containing the ephemeral key $R$ and the viewing meta-address $P_{\text{view}}$ can never be erased from past blocks. An adversary equipped with a fault-tolerant quantum processor in the 2030s could run Shor's algorithm against $R$ or $P_{\text{view}}$, recover the private viewing scalar $v_{\text{view}}$, recompute every historical shared secret $S = R \cdot v_{\text{view}}$, and retroactively deanonymize years of private financial operations conducted by enterprises and individuals.
Hybrid secp256k1 + ML-KEM architecture for everlasting Web3 privacy
To eliminate retroactive deanonymization without breaking backward compatibility with existing EVM smart contracts, the working group led by the Ethereum Foundation, Umbra, and Fluidkey engineered a hybrid stealth address derivation scheme that pairs elliptic curve cryptography with the post-quantum ML-KEM-768 standard (alongside a compact ML-KEM-512 variant optimized for Layer 2 rollups with strict calldata constraints).
In this hybrid architecture, a user's viewing meta-address contains both an elliptic curve point $P_{\text{view}}$ and a lattice public encapsulation key $ek_{\text{view}}$. When constructing a private transfer, the sender's wallet simultaneously executes the classical ECDH key agreement to obtain $S_{\text{ecc}}$ and runs ML-KEM.Encaps(ek_view) to produce a post-quantum shared secret $S_{\text{pq}}$ along with its encapsulated ciphertext $ct$. Both secrets are fused through a Keccak-256 key derivation function bound to the network identifier:
$$S_{\text{hybrid}} = \text{keccak256}(S_{\text{ecc}} \parallel S_{\text{pq}} \parallel \text{chainId})$$
The ML-KEM ciphertext $ct$ is published inside the metadata field of the Announcement event (or inside low-cost EIP-4844 blobs on Layer 2 rollups such as Arbitrum, Optimism, Base, and Starknet), accompanied by a 1-to-2-byte viewTag derived from $S_{\text{hybrid}}$. Because ML-KEM decapsulation relies on the computational hardness of the Module Learning With Errors (Module-LWE) lattice problem, no future quantum computer will be able to extract $S_{\text{pq}}$ from the ciphertext $ct$ archived on-chain.
On an immutable public blockchain, privacy that relies solely on elliptic curves carries an expiration date; only post-quantum key encapsulation guarantees that today's financial graph remains confidential thirty years from now.
This cryptographic roadmap aligns directly with broader industry shifts toward post-quantum algorithms in enterprise networks and complements the physical enclave protections detailed in our analysis of hardware wallet security against side-channel attacks.
| Technical parameter | Classical ERC-5564 (Scheme ID 1) | New Hybrid Stealth Proposal (EF + Umbra + Fluidkey + Trezor) | Cryptographic and operational benefit |
|---|---|---|---|
| Shared secret exchange primitive | Pure secp256k1 ECDH |
Hybrid secp256k1 + ML-KEM-768 / ML-KEM-512 (FIPS 203) |
Complete immunity against retroactive quantum deanonymization |
| On-chain announcement payload | 33 bytes (compressed ephemeral key) + 1-byte viewTag |
33 bytes ECC + 768/1088 bytes ct (optimized via L2 blobs) |
Low-cost storage on rollups with state compression |
| Fast scanning filter | 1-byte viewTag (skips 255/256 of non-matching logs) |
2-byte viewTag + isolated viewing key delegation |
Instant mobile scanning without exposing spending keys |
| Hardware wallet support | Limited by non-standard ad-hoc key tweaking | Native Trezor firmware integration with domain separation | Cold storage custody for assets received at stealth addresses |
Trezor hardware wallet integration and fast-scanning view tag architecture
One of the steepest historical barriers to widespread stealth address adoption across Web3 has been friction with hardware wallets. To spend funds received at a stealth address $P_{\text{stealth}}$, a hardware signing device must add the scalar tweak $\text{keccak256}(S_{\text{hybrid}})$ to the master spending private key $v_{\text{spend}}$ inside its protected memory—an operation that standard BIP-32/BIP-44 derivation flows did not natively support.
Bringing Trezor into the Ethereum Foundation working group resolves this bottleneck through a formal firmware command specification for hardware signing devices. Under this model, the private viewing key ($d_{\text{view}}$ for ML-KEM and $v_{\text{view}}$ for the elliptic curve) is derived along a cryptographic branch completely segregated from the spending key ($v_{\text{spend}}$). Users can authorize their web or mobile interface (such as Fluidkey or the Umbra app) to use the exported viewing key to continuously scan blockchain event logs and filter via the viewTag which incoming transfers belong to them.
Whenever the user wants to transfer or consolidate assets sitting in a stealth address, the host application passes only the verified masking scalar to the Trezor device. The hardware wallet firmware validates the ownership proof, computes the ephemeral spending private key in volatile RAM without ever exposing the root seed to the connected computer, and signs the transaction (or the ERC-4337 Account Abstraction UserOperation). To verify the integrity of deployed registry contract bytecodes and firmware binaries, developers can use a cryptographic hash generator alongside formal SHA-256 file integrity verification workflows, while auditing RPC gateway endpoints with an SSL certificate inspector.
TypeScript reference implementation for hybrid ML-KEM stealth derivation
The following TypeScript code block demonstrates the cryptographic workflow executed by a sending client to encapsulate a hybrid shared secret (secp256k1 + ML-KEM-768), compute the 2-byte fast-filtering viewTag, and derive the one-time Ethereum stealth destination address:
import { secp256k1 } from '@noble/curves/secp256k1';
import { ml_kem768 } from '@noble/post-quantum/ml-kem';
import { keccak_256 } from '@noble/hashes/sha3';
import { bytesToHex, concatBytes, utf8ToBytes } from '@noble/hashes/utils';
interface HybridStealthMetaAddress {
spendingPubKey: Uint8Array; // 33 bytes (compressed secp256k1)
viewingEccPubKey: Uint8Array; // 33 bytes (compressed secp256k1)
viewingMlkemPubKey: Uint8Array; // 1184 bytes (ML-KEM-768 encapsulation key)
}
export function generateHybridStealthAddress(
meta: HybridStealthMetaAddress,
chainId: bigint = 1n
) {
// 1. Classical ECDH key agreement over secp256k1 using an ephemeral key pair
const ephemeralPrivKey = secp256k1.utils.randomPrivateKey();
const ephemeralPubKey = secp256k1.getPublicKey(ephemeralPrivKey, true);
const sharedEccSecret = secp256k1.getSharedSecret(
ephemeralPrivKey,
meta.viewingEccPubKey,
true
);
// 2. Post-quantum ML-KEM-768 key encapsulation (FIPS 203)
const { cipherText, sharedSecret: sharedPqSecret } = ml_kem768.encapsulate(
meta.viewingMlkemPubKey
);
// 3. Derive the chain-bound hybrid shared secret
const domainTag = utf8ToBytes(`ERC5564-HYBRID-MLKEM768-CHAIN-${chainId.toString()}`);
const hybridSecret = keccak_256(
concatBytes(sharedEccSecret, sharedPqSecret, domainTag)
);
// 4. Extract a 2-byte viewTag for ultra-fast scanning (1 in 65,536 false positive rate)
const viewTag = bytesToHex(hybridSecret.slice(0, 2));
// 5. Compute the stealth public key point: P_stealth = P_spend + h*G
const tweakScalar = BigInt('0x' + bytesToHex(hybridSecret)) % secp256k1.CURVE.n;
const SpendPoint = secp256k1.ProjectivePoint.fromHex(meta.spendingPubKey);
const TweakPoint = secp256k1.ProjectivePoint.BASE.multiply(tweakScalar);
const StealthPoint = SpendPoint.add(TweakPoint);
const uncompressedStealth = StealthPoint.toRawBytes(false).slice(1);
const stealthAddress = '0x' + bytesToHex(keccak_256(uncompressedStealth).slice(-20));
return {
stealthAddress,
ephemeralPubKey: '0x' + bytesToHex(ephemeralPubKey),
mlkemCipherText: '0x' + bytesToHex(cipherText),
viewTag: '0x' + viewTag,
};
}
Standardization pillars defined by the working group
The technical roadmap published by the Ethereum Foundation alongside Umbra, Fluidkey, and Trezor structures the rollout across three open-source deliverables:
- New scheme identifier (
schemeId) for ERC-5564: formal definition of thesecp256k1-mlkem768hybrid profile inside the ERC-6538 meta-address registry to guarantee seamless cross-wallet interoperability. - Private gas sponsorship via ERC-4337 Paymasters: native integration with Account Abstraction so a recipient can withdraw ERC-20 tokens or stablecoins from a freshly funded stealth address without having to pre-fund it with ETH from a linkable external wallet to pay gas fees.
- Announcement compression on Layer 2 rollups: optimized packing of ML-KEM ciphertexts utilizing precompiles and ephemeral blob storage to keep the cost of each private transfer below one cent on L2 networks.
Default, future-proof privacy for the Ethereum ecosystem
The official launch of this working group demonstrates that the Ethereum ecosystem is tackling on-chain privacy and the quantum threat as a single, inseparable architectural challenge. By combining the stealth payment protocol expertise of Umbra and Fluidkey, the silicon hardware custody engineering of Trezor, and the Ethereum Foundation's cryptographic research on ML-KEM, Web3 moves toward a unified standard where enterprises and individuals can transact on public ledgers with mathematical certainty that their financial confidentiality will withstand both modern graph analytics and future quantum computation.


