Ethereum Faces a 52-Bit Security Gap in Its zkEVMs

Ethereum has already won the speed battle on zkEVMs. The security one remains open. A research contest supported by the Ethereum Foundation still shows a gap of 52.14 bits between the currently proven results on a cryptographic component. The network targets a demonstrable security of 128 bits for its zkEVMs by early December. No changes threaten the current consensus for now.

Ethereum wants 128 bits of security

Ethereum has been preparing for a long time the arrival of zkEVMs directly on its main layer. The work initially focused on performance. By the end of 2025, the teams had managed to reduce the time required to prove an Ethereum block from about 16 minutes to a few seconds. Several zkVMs could already generate proofs for 99% of the blocks in less than ten seconds on the targeted hardware.

Speed is no longer enough. The Ethereum Foundation now demands provable security of 128 bits. The teams will also have to produce final proofs of 300 KB maximum and precisely document how their different proof systems work together.

The next major deadline is early December. Before that, researchers still have to improve several cryptographic building blocks. One of them is now under the spotlight.

A gap of 52.14 bits remains open

The site better.codes organizes a contest where researchers can try to improve or attack a parameter called koalaIRS12. The dashboard currently displays two figures. The lower security bound reaches 63.99 bits. The upper bound drops to 116.13 bits. Between them: 52.14 bits still unresolved.

These values do not mean that Ethereum today operates with only 63.99 bits of security. The benchmark measures a precise cryptographic profile used in research on proof systems. Not the entire network.

This precision matters since Ethereum wants to make Zero-Knowledge a much larger part of its architecture. The contest works both ways. Some researchers try to raise the security bound by providing new mathematical proofs. Others look for attacks to lower the upper limit.

Nine results from seven researchers had already been accepted by August 21. The goal is to gradually reduce the space between the two. Full zkEVMs will require more. Different components, recursion of proofs, and the match between mathematical models and the actual code used will still have to be verified. December is approaching.

The current Ethereum network does not yet depend on these proofs

The tight schedule does not mean that a flaw threatens Ethereum transactions today. The zkEVMs concerned do not yet replace the normal consensus operation. In the current phase, their proofs can accompany mainnet tests. Classic execution clients continue to recalculate transactions before validators attest to the blocks.

Ethereum therefore still operates as usual. The situation will change if zk proofs one day become reliable enough to replace part of this re-execution. This is precisely why the Ethereum Foundation insists so much on 128 bits.

A false proof accepted at layer 1 could have much more serious consequences than a bug in an application. It could theoretically allow validating an incorrect network state. The protocol is already multiplying efforts around this subject. In July, Ethereum notably deployed AI agents to automatically search for critical vulnerabilities. zkEVMs now follow the same principle. Ethereum already knows how to prove quickly. What remains is to prove without leaving any doubt.