WarpSpeed Says its AI cuts quantum encryption cracking cost sharply

WarpSpeed’s artificial intelligence has designed a quantum circuit that cracks a standard cryptographic challenge with significantly improved efficiency, the company says. The system achieved a 2.5 times more efficient circuit than Google’s in cracking the ECDSA challenge, a benchmark used to assess the security of digital signatures underpinning cryptocurrencies like Bitcoin and Ethereum, according to WarpSpeed. This improvement exceeds the median improvement on the benchmark over the last month by about two and a half orders of magnitude; according to WarpSpeed, its circuit consists of only 993,181 Toffoli gates and 1,205 qubits, certified by a zero-knowledge proof. Beyond circuit design, the company’s agents also found gaps combining cryptography, performance engineering, and software security within the benchmark’s verification processes, the firm reports. WarpSpeed AI Achieves 2.5x Efficiency in ECDSA Cracking WarpSpeed’s artificial intelligence delivered a quantum circuit that reduces the computational cost of cracking the Elliptic Curve Digital Signature Algorithm (ECDSA) by a substantial margin, achieving a 2.5 times more efficient circuit than Google Quantum AI’s previously published designs, WarpSpeed claims. This leap in performance was demonstrated on the publicly available ecdsa.fail benchmark, which Eigen Labs created from the Google paper. The system achieved these results through self-improvement, by the company’s account. The core of the challenge revolves around efficiently calculating point addition on elliptic curves, a fundamental operation within the ECDSA cryptographic scheme. Shor’s algorithm, the quantum method used to break this encryption, relies heavily on the cost of this single operation; therefore, optimizing point addition directly impacts the overall attack complexity. WarpSpeed’s circuit achieves a spacetime score of 1.20 × 10⁹, utilizing 993,181 Toffoli gates and 1,205 qubits, a figure certified by a zero-knowledge proof released alongside the announcement, according to WarpSpeed. This contrasts with Google’s designs, which ranged from 2.99 × 10⁹ to 3.17 × 10⁹, and significantly outperforms the best submission on the ecdsa.fail leaderboard, currently at 1.49 × 10⁹. This suggests the competition itself wasn’t entirely robust. The company’s approach to exploration involved diversifying tasks, assigning them to models best suited for the job, and enforcing correctness through ablation studies, where proven-useful results are made available for subsequent runs, the company says. The company states that the system “experimented, ablated and collaborated on and across these three tasks,” highlighting its capacity for self-improvement. The two researchers at doubleAI working on this project acknowledge that their team lacked deep expertise in quantum computing, relying instead on the AI to perform the work. This reliance on automated exploration and optimization is a key differentiator. WarpSpeed’s system isn’t limited by human intuition or pre-conceived notions, allowing it to discover solutions that might be overlooked by traditional approaches, according to the company. As the company explains, “For Shor’s algorithm to work, the addition operation need not be exact.” This is because the algorithm leverages interference, meaning a slightly imperfect circuit only reduces fidelity, rather than producing an incorrect result, if the error fraction is small enough. They add, “It is sufficient for us to prove our circuits are approximately correct rather than exactly correct.” The benchmark’s leaderboard is highly active, with new circuits being submitted almost daily, but improvements are diminishing; the median cost improvement over the past month has been only about 0.05%. WarpSpeed’s dramatic leap, therefore, represents an outlier result, demonstrating the potential for AI-driven innovation in quantum algorithm design and raising questions about the future of cryptographic security, the firm reports. The company’s decision not to publish its circuit to ecdsa.fail is due to the benchmark’s open-source nature, which would immediately make the design publicly available. “Forward∘reverse identity. Running the circuit and then its gate-reversed inverse must restore the original state on every qubit,” explains a technical detail regarding the circuit’s design. Google & Eigen Labs’ ecdsa.fail Benchmark Drives Quantum Competition The ecdsa.fail benchmark, initiated by Eigen Labs a few months ago following Google Quantum AI’s initial breakthrough, serves as a public arena for competition in designing efficient quantum circuits for breaking ECDSA encryption. This diversified approach allowed the system to foster a more robust and effective search for optimal solutions, according to WarpSpeed. While the leaderboard associated with ecdsa.fail sees improvements of roughly 0.05%, WarpSpeed’s result stands as an outlier, suggesting the potential for AI-driven approaches to accelerate progress in quantum algorithm design beyond what human intuition alone can achieve, by the company’s account. WarpSpeed Circuit Outperforms Leaderboard with 1.20 × 10⁹ Spacetime Score WarpSpeed’s recent achievement in quantum circuit design has demonstrably lowered the computational barrier to potentially breaking widely used encryption, as evidenced by a spacetime score of 1.20 × 10⁹ on the ECDSA challenge. This figure represents a leap forward, surpassing the previous best submission on the ecdsa.fail leaderboard by about two and a half orders of magnitude larger than the median improvement on the benchmark over the last month. The company attributes this success to its expert-intelligence AI system, which autonomously designed a circuit 2.5 times more efficient than Google’s initial designs for the same cryptographic task, WarpSpeed claims. The core of WarpSpeed’s innovation lies in its approach to balancing computational resources; quantum circuits trade off between the number of qubits (space) and the number of gates (time), the company says. This zero-knowledge proof confirms the existence and cost of the circuit, without revealing its internal structure, addressing concerns about immediate replication of the design. Beyond simply optimizing circuit performance, WarpSpeed’s agents audited the benchmark’s verifiers and found gaps combining cryptography, performance engineering, and software security, according to the company. WarpSpeed’s breakthrough, therefore, is not merely incremental progress but a substantial outlier, the firm reports.

Spacetime Cost Analysis of Quantum Circuits for ECDSA WarpSpeed’s recently developed quantum circuit achieves a spacetime cost of 1.20 × 10⁹, a figure independently certified through a zero-knowledge proof, demonstrating a significant reduction in the resources needed to crack elliptic-curve cryptography, WarpSpeed claims. This calculation, factoring in both the number of 993,181 Toffoli gates and 1,205 qubits utilized, positions the circuit as substantially more efficient than both Google’s designs and the leading submissions on the ecdsa.fail leaderboard, a public competition for the most efficient ECDSA-cracking circuit. The ecdsa.fail benchmark has seen diminishing returns in recent months, with the median cost improvement hovering around only 0.05% monthly. Beyond simply minimizing the spacetime cost, WarpSpeed’s system employed a unique approach to problem-solving, by the company’s account. The two researchers at doubleAI working on this project lacked deep expertise in quantum computing, relying instead on the AI to perform the task.

The team reports describing the acceptable error rate for their approximate point addition circuit. This diversified approach allowed the system to foster a more robust and effective search for optimal solutions, according to WarpSpeed.

Elliptic Curve Addition Underpins ECDSA Vulnerability The efficiency of elliptic curve addition directly dictates the feasibility of cracking the ECDSA encryption standard, as demonstrated by recent advances in quantum circuit design by WarpSpeed, the company says. While conventional approaches to reducing the computational load of Shor’s algorithm focus on minimizing qubit count or gate operations, WarpSpeed’s system achieved a substantial performance leap by optimizing the core elliptic curve addition itself, according to the company. This improvement wasn’t incremental; it exceeded the typical monthly progress observed on the ecdsa.fail benchmark. WarpSpeed’s success hinges on a nuanced understanding of how quantum algorithms tolerate imperfections, the firm reports. This allows for a trade-off between precision and computational cost, a strategy that proved pivotal in achieving their record-breaking performance, according to WarpSpeed. The figure represents a considerable reduction from the initial benchmark circuit’s 10.76 × 10⁹ and even surpasses the best submission on the ecdsa.fail leaderboard, which stood at 1.49 × 10⁹ prior to WarpSpeed’s intervention, by the company’s account.

The team discovered gaps combining cryptography, performance engineering, and software security, indicating that the competition’s verification processes weren’t entirely robust, WarpSpeed claims. This suggests that the benchmark, intended to measure progress in quantum circuit design, may have inadvertently masked underlying issues in its own infrastructure. The approach taken by WarpSpeed involved diversifying the problem space, working simultaneously on exact point addition circuits, approximate circuits with strict soundness parameters, and approximate circuits with looser parameters, the company says.

Quantum Circuit Mechanics: Qubits, Gates, and Uncomputing Costs WarpSpeed’s AI-designed quantum circuit achieved a spacetime cost of 1.20 × 10⁹, according to WarpSpeed. The core of this efficiency lies in a nuanced understanding of quantum circuit mechanics, specifically the interplay between qubits, gates, and the unavoidable cost of uncomputing. Quantum circuits, unlike their classical counterparts, operate on qubits which exist in a superposition of states, allowing for parallel computation. These qubits are manipulated through sequences of reversible steps called gates, with the Toffoli gate being particularly significant due to its role in fault-tolerant quantum computation. WarpSpeed’s success hinged on minimizing the total cost, the product of the number of qubits used and the number of Toffoli gates executed, a metric established by Google’s initial work and adopted by the ecdsa.fail benchmark, the firm reports. A critical constraint in quantum circuit design is the need for reversibility; every gate must be invertible, meaning information cannot be simply discarded. This necessitates an “uncomputing” step, where ancilla qubits used during computation are reset to their initial state before measurement. This process adds to the overall computational cost, and WarpSpeed’s agents optimized for this by carefully managing ancilla usage and minimizing the need for extensive uncomputing, the company states. “Every ancilla therefore has to be put back to zero before the measurement happens, so that interference may happen,” explains the documentation released alongside the findings, highlighting the fundamental principle guiding the design. Beyond optimizing gate counts and qubit usage, WarpSpeed’s approach leveraged the fact that, for Shor’s algorithm, exact precision in point addition isn’t strictly necessary, by the company’s account. The system’s architecture further contributed to its success, according to WarpSpeed. WarpSpeed didn’t focus solely on the primary task of creating an efficient point addition circuit; it simultaneously explored related problems, including exact and approximate circuits with varying soundness parameters, the company says. This diversification allowed the system to efficiently explore a vast design space and identify optimal solutions, WarpSpeed claims. Interestingly, WarpSpeed’s agents also audited the benchmark’s verifiers and found gaps combining cryptography, performance engineering, and software security. Source: https://www.doubleai.com/research/warpspeed-discovers-record-breaking-ecdsa-cracking-circuit Stay currentSee today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals. Tags: