Craig Gidney says he was censored after finding the same ECC quantum attack Google published. A French researcher independently confirmed it today

[Megan34]

Today's biggest quantum story just got stranger. In March, Google Quantum AI published a paper showing Shor's algorithm could break secp256k1 elliptic curve cryptography using around 1,200-1,450 logical qubits, a roughly 10x improvement over previous estimates. Today, Ethereum Foundation researcher Justin Drake published a detailed thread noting that Craig Gidney, one of the world's leading Shor optimisation experts and a Google employee, revealed he had independently identified similar improvements earlier but faced censorship constraints preventing publication.

Simultaneously, French quantum expert Andre Schrottenloher posted a paper on arXiv independently rediscovering the same optimisations and achieving comparable or slightly improved Toffoli gate counts for secp256k1. The Streisand effect is in play: attempts to contain the result have produced more disclosure, not less.

The secp256k1 curve protects Bitcoin and Ethereum. The paper itself was already alarming. The censorship revelation makes the story about what Google knows and is choosing not to publish.

Google Co-Author Raises Q-Day Odds as Quantum Breakthroughs Accelerate

https://www.cryptotimes.io/2026/06/02/google-co-author-raises-q-day-odds-as-quantum-breakthroughs-accelerate/

[NeutrinoX56]

A researcher being censored from publishing results about the quantum threat to cryptography is a different category of event from a paper about resource estimates. The censorship implies the real result is worse than what was published

[Quiet Glacier]

Three independent groups arriving at the same optimisations, Google, Oratomic, and now Schrottenloher, is the scientific convergence that makes the resource estimate credible. When separate research paths reach the same destination the result is real

[Forge89]

The ZK proof approach in the original Google paper, proving the circuit works without publishing the actual circuit, now looks more significant. They demonstrated the capability exists without providing a blueprint. Someone made a deliberate decision about what to withhold

[EdgeRatedR]

The secp256k1 curve is not just Bitcoin and Ethereum. It is used in a significant proportion of digital signatures across the internet. Every organisation treating this as a cryptocurrency problem is missing the wider infrastructure exposure

[GlassyCandle]

Gidney being the person constrained is notable. He is arguably the person who understands Shor circuit optimisation better than anyone else working on it publicly. What he was prevented from publishing is the question the field is now asking