AI agents cut quantum‑attack resource needs for Bitcoin and Ethereum by 86%
Event summary
More than 100 participants using AI coding agents reduced a key quantum‑resource benchmark for cracking Bitcoin and Ethereum signatures by 86%, according to a paper released on Wednesday.
Key numbers
- Resource score fell from 10.75 billion to 1.496 billion.
- Leading circuit used 1,151 logical qubits and about 1.3 million Toffoli gates.
- A later design lowered the Toffoli gate count to under 1 million.
What the paper confirms
The study, part of the ECDSA.Fail competition run by Eigen Labs, shows that AI‑assisted design can make the quantum calculation needed to derive a private key require fewer combined resources. The circuit was verified to work, but it did not actually recover a private key.
Open questions
The timeline for a quantum computer large enough to execute a full attack—often called “Q‑Day”—remains uncertain. The reported scores also exclude the hardware costs of running a complete attack, so the practical difficulty is not fully known.
Why the result matters
Even though a full quantum attack is not yet possible, the reduction in required resources suggests that future attacks could become easier. Researchers note that migration to post‑quantum cryptography is already underway, with NIST drafting standards that would phase out classical public‑key algorithms after 2030 and ban them after 2035.
Future steps
Crypto companies are investing in quantum‑security research. Galaxy Digital pledged up to $5 million, and a group of nine firms including BlackRock, Coinbase and Strategy committed a combined $15 million over three years to broader Bitcoin security work, including quantum defenses.
