Loading story…
Trail of Bits disclosed that it forged a zero-knowledge proof accepted by Google Quantum AI’s unpatched verifier for a published quantum-circuit claim tied to elliptic curve cryptanalysis. The researchers said the result did not break the underlying cryptography or demonstrate a new quantum algorithm; instead, it exploited weaknesses in Google’s Rust-based prover and simulator, including unsafe deserialization of untrusted circuit bytes and logic flaws that let invalid operation types evade Toffoli-gate accounting and use register aliasing to violate reversibility constraints.
According to Trail of Bits, the forged proof reused the same verification key and reported stronger-looking metrics than Google’s published proof, including 8,288,880 operations, 1,164 qubits, and 0 Toffoli gates, while actually embedding arbitrary classical logic into what was supposed to be a quantum circuit model. The firm published a proof-of-concept repository with the forged artifact, verification steps, a circuit-generation script, and a patch affecting code outside the zkVM; it said the proof was generated on a cloud system with four NVIDIA H100 GPUs in about four hours. Google patched the proof implementation and said its underlying scientific claims were unaffected.
See real exploitation activity before you spend the cycle.
6 events from the most recent confirmed update back to the earliest known activity.
Trail of Bits published a blog post and GitHub repository describing how it beat Google's zero-knowledge proof of quantum cryptanalysis by exploiting implementation vulnerabilities rather than achieving a quantum algorithm breakthrough.
By the time of disclosure, Google had patched the proof implementation in version 2 of the verification workflow. Google stated that the implementation issues did not affect the underlying scientific claims.
Trail of Bits created a proof-of-concept repository containing the forged proof artifact, verification instructions, a circuit-generation script, and a patch manipulating kickmix circuit serialization. The firm said it generated the forged proof on a cloud system with four NVIDIA H100 GPUs in about four hours.
Using the identified flaws, Trail of Bits produced a forged zero-knowledge proof that reportedly verified with the same verification key as Google's published proof while claiming stronger metrics, including about 8.3 million operations, 1,164 qubits, and zero Toffoli gates.
Trail of Bits found multiple vulnerabilities in the unpatched proof system implementation, including unsafe deserialization of untrusted circuit bytes and logic flaws that let invalid operations bypass Toffoli accounting and violate reversibility constraints through register aliasing.
Before Trail of Bits' disclosure, Google Quantum AI had published a proof-verification workflow and proofs supporting claims about a quantum circuit for elliptic curve cryptanalysis. Trail of Bits' later analysis indicates the attacked target was version 1 of this workflow.
Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.
See real exploitation activity behind this advisory so you can triage it against everything else in the queue.
2 references tracked. Mallory keeps watching after this page renders.
blog.trailofbits.com
Open sourcegithub.com
Open sourceMap indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
Post-quantum security researchers warned that public reporting is becoming an unreliable measure of progress toward cryptographically relevant quantum computers, after high-profile disclosures on Shor-based attacks against elliptic-curve cryptography exposed tensions between openness and national-security concerns. Oratomic publicly released resource estimates and a responsible-disclosure statement for fault-tolerant quantum cryptanalysis, while Google published `ECDLP-256` resource estimates only after consulting the U.S. government and withheld the underlying circuits behind a zero-knowledge proof. The disclosures were presented as evidence that publication decisions around quantum cryptanalysis are becoming sensitive in their own right, with U.S. export controls, executive actions, and other intervention mechanisms cited as tools that could further limit what becomes public. Google’s partial-disclosure approach quickly unraveled: André Schrottenloher independently reconstructed the concealed circuits, and Trail of Bits reportedly forged a proof by exploiting bugs in Google’s prover code, undermining the attempt to reveal results without fully exposing implementation details. Commentaries tied those events to a broader warning that future breakthroughs may be selectively disclosed or kept secret altogether, echoing older intelligence-era norms around cryptanalytic advances. For defenders, the practical message was to stop using public quantum milestones as the trigger for post-quantum migration and instead treat published capability as a lower bound while accelerating cryptographic inventory, dependency mapping, and migration planning.
Jul 12, 2026
Google Quantum AI reported that breaking the `secp256k1` elliptic-curve cryptography used by Bitcoin, Ethereum, and other cryptocurrencies may require far fewer quantum resources than previously estimated. In a new whitepaper co-authored with researchers from the Ethereum Foundation, Stanford, and UC Berkeley, the team said an attack could need fewer than 500,000 physical qubits rather than the millions often cited, with optimized circuits using roughly 1,200 to 1,450 high-quality logical qubits. Under idealized conditions, the research suggests a primed attacker could recover a private key in about nine minutes, making an on-spend attack against a Bitcoin transaction theoretically possible within the network’s typical 10-minute confirmation window. The paper warns that Bitcoin’s 2021 **Taproot** upgrade may have increased long-term quantum exposure by making public keys visible on-chain by default, and estimates that about **6.9 million bitcoin** are already held in wallets with exposed public keys. Google said it did not publish the attack circuits, instead releasing a zero-knowledge proof using **SP1 zkVM** and **Groth16** so the results could be validated without providing weaponizable details, and said it notified the U.S. government before publication. While the researchers stressed that no cryptographically relevant quantum computer exists today, they argued that the engineering threshold for attacking ECC-based blockchain systems is materially lower than expected and called for migration toward post-quantum cryptography across cryptocurrency ecosystems.
Jun 29, 2026
IBM, the University of Chicago, and several partners published three preprint papers describing quantum-computing experiments on **IBM Heron** processors that aim to show computations beyond leading classical simulation methods while improving confidence in the correctness of the results. The most prominent paper introduced **doped Clifford sampling**, combining Clifford circuits, spacetime-code error detection, and injected `T` gates to create a classically hard computation with a built-in fidelity certificate; the team reported a 70-qubit circuit encoded across 97 physical qubits and a fidelity lower bound of **0.284** with **95% confidence**. Two additional studies, including work involving **Qedma**, **RIKEN**, **BlueQubit**, and **Algorithmiq**, reported quantum simulation results that remained stable after tested classical methods stopped converging or produced inconsistent outputs, using techniques such as embedded error checks, postselection, error mitigation, repeat runs, and cross-platform comparisons with trapped-ion systems. The papers have not been peer reviewed, do not rule out future classical algorithmic advances, and have not demonstrated commercial utility, but they are being presented as a significant step toward more credible and verifiable claims of quantum advantage.
Aug 2, 2026