The Ledger
Ethereum Draft Pitches BLS48-581 as a Backup for Validator Signatures
A November 2 Ethereum draft proposes registering dormant BLS48-581 keys for consensus-critical validator signatures, while leaving several duties on BLS12-381. The proposal targets a hypothetical classical cryptographic attack, but its author says client performance and key-registration questions remain unresolved.

Ethereum’s draft proposal for larger validator-signature keys aims to prepare for a cryptographic attack that researchers have not observed, while asking node operators to absorb a costly curve change that has not yet been tested in a client. The November 2 draft names BLS48-581 as its current candidate to replace BLS12-381 for Ethereum’s most security-critical consensus signatures. It also warns that the new curve would bring much larger signatures and higher verification costs.
Source gist.github.com
A hedge against a break that has not happened
Author Saulius Grigaitis frames the proposal around a hypothetical classical attack that halves elliptic-curve security. The draft estimates BLS12-381 at roughly 117 to 126 bits of security today, compared with a target of 256 bits for BLS48-581; under the proposed threat model, the latter would retain about 128 bits. The document says no attack of this kind is known, and cautions that a real attack might affect the group and field components of a pairing differently.
“This EIP moves the most security-critical consensus-layer BLS signatures from BLS12-381 to a higher-security pairing-friendly curve, so that they would survive a future attack that cuts elliptic curve security in half,” Grigaitis writes in the draft. The proposal does not offer protection against a sufficiently powerful quantum computer: it says Shor’s algorithm would break both curves. The larger curve is presented as a hedge against a possible classical advance, not a post-quantum solution.
Two keys, with a later fork to switch signing
The design would register a new BLS48-581 signing key alongside validators’ post-quantum keys through EIP-8450. The key would remain dormant until a later fork activates it; that signing fork could be an emergency hard fork, and the chain could instead move to post-quantum keys if those are ready and judged appropriate. The draft has no EIP number or pull request, and lists no external reviews as of November 2.
Keeping each validator’s existing 48-byte BLS12-381 public key as its permanent identifier avoids changing the deposit contract and execution-layer request formats, according to the draft. The proposed BLS48-581 public key is 73 bytes, while a signature in the selected G2 group would be 584 bytes, up from 48-byte keys and 96-byte signatures on the existing curve. Validators would need both keys online after the switch, and clients would have to keep both curves available to verify signatures from before and after the fork.
Not every BLS duty would move
The draft would shift signatures used for attestations, block proposals, voluntary exits, sync committees, payload attestations, builder bids, proposer preferences and inclusion lists. It leaves RANDAO reveals, selection proofs, aggregator signatures, deposits and BLS-to-execution changes on BLS12-381. The author’s proposed distinction is whether a forged signature could directly forge votes, blocks or other consensus actions; the retained duties, the draft argues, do not allow that same form of forgery.
Blobs also remain outside the proposal. Moving their KZG commitments would require a new trusted setup, and the draft says BLS48-581’s largest power-of-two evaluation domain is 210, compared with the 8192-element domain PeerDAS requires. The document points to calldata as a fallback for blob users if BLS12-381 were broken, though it notes that calldata has less capacity and costs more.
Performance and readiness are unresolved
The draft’s initial modeling suggests the change could work on ordinary multi-core computers for nodes on default attestation subnets, provided implementations use batching and keep the proposed signature split. It has not been demonstrated in a running client, and nodes subscribed to every attestation subnet may need more powerful hardware, optimized libraries or lower attestation load. Larger signatures and slower verification could also put pressure on gossip processing and attestation deadlines.
Several design choices remain open, including the final curve, whether signatures should use the G1 or G2 group, how to handle validators without registered keys at the switch, and whether one-way signing keys need a recovery path. BLS48-581 is therefore a candidate, not a settled protocol choice. Before any switch, the draft says, validators would need to register keys in advance and client teams would need to test implementations, because keys registered after a break in BLS12-381 could no longer be trusted.
Sources
- Original source GitHub Gist gist.github.com
Other pages read
- Fellowship of Ethereum Magicians ethereum-magicians.org