Independent Labs Break Google’s Secret Crypto Code

In March, a major breakthrough emerged from Google scientists and their collaborators that brought the possibility of a quantum computer capable of breaking today’s internet encryption significantly closer to reality. While the cybersecurity community typically shares attack methodologies to enable better defenses, this particular group took an unprecedented step. They believed their discovery posed such a serious security threat that they deliberately concealed the specifics of how to replicate their work.

Remarkably, within just three days, the Seattle-based research startup Eigen Labs not only matched those hidden results but exceeded them, leveraging crowdsourcing and swarms of AI agents. The theoretical power of quantum computers lies in their ability to solve problems that would take classical computers millennia to crack. Modern cryptography relies on the difficulty classical machines have with tasks like factoring enormous numbers, but using an algorithm developed by mathematician Peter Shor in 1994, quantum computers could theoretically dismantle such encryption rapidly.

Although no practical quantum code-breaking hardware exists today, laboratories worldwide are racing to build machines with enough qubits,the quantum version of classical bits,to execute these attacks. This looming threat has already prompted governments to begin transitioning to post-quantum cryptography (PQC). In the U. S., federal agencies must move high-value assets and high-impact systems to PQC for key establishment schemes by the end of 2030. Experts argue that the findings from Google and Eigen Labs make it clear: the migration to quantum-resistant encryption must happen as quickly as possible.

Preparing for Post-Quantum Cryptography

To prepare for the era of cryptographically relevant quantum computers, scientists are constantly probing the resources such machines would require. In 2025, Google Quantum AI researcher Craig Gidney demonstrated that a quantum computer with fewer than 1 million qubits, running Shor’s algorithm for less than a week, could break 2,048-bit RSA encryption,a common standard for securing online data. This represented a 20-fold reduction in qubit requirements compared to 2019 estimates.

Gidney and his team then turned their attention to elliptic curve cryptography (ECC), the encryption that underpins the security of cryptocurrencies like Bitcoin and Ethereum, and alongside RSA, protects most modern internet traffic. On March 30, they announced they had optimized Shor’s algorithm to break 256-bit ECC using just 1,200 to 1,450 logical qubits. (A logical qubit is composed of many error-prone physical qubits; current state-of-the-art processors, like IBM’s Condor with 1,121 qubits, are still far from this threshold.) The researchers calculated that these computations could be encoded with fewer than 500,000 superconducting physical qubits, cracking 256-bit ECC in 18 to 23 minutes,again, a nearly 20-fold reduction in physical qubit estimates.

“I knew we could do better but was not expecting that much improvement,” said David Jao, a professor at the University of Waterloo who was not involved in the study. However, instead of fully explaining their method, the scientists published their work using a zero-knowledge proof,a technique that verifies an attack works without revealing how to execute it. “To my knowledge, this was the first time that a result of this kind was released using a zero-knowledge proof,” said André Schrottenloher, a researcher at the Inria Center at the University of Rennes.

Google stated in a blog post that it concealed its results after discussions with the U. S. government. Yet most experts questioned the necessity. Steven Galbraith, a professor at the University of Auckland, called it “a cute way to use a zero-knowledge proof” but doubted that cryptographically relevant quantum computers are imminent. Others were more critical. “Zero-knowledge proofs for academic research are both useless and futile,” Jao said. “The purpose of academic research is not merely to answer questions, but to inform the community and allow other teams to build upon the results. A zero-knowledge proof does not convey understanding.”

Replicating Google’s Results

At Eigen Labs, 22-year-old engineer Gautham Anant, enrolled in an introductory quantum computing course at the University of Washington, decided to see if he could replicate Google’s work. By analyzing the virtual machine Google built to verify its findings, Anant created software to test any quantum circuit’s efficiency in breaking 256-bit ECC. With help from fellow engineer Gajesh Naik, he deployed AI agents to scan scientific literature and automatically design and optimize quantum circuits.

Initially, Eigen Labs could not match Google’s efficiency. So on June 1, they launched a public site where anyone could direct their own AI agents to improve the circuits, with agents sharing notes and building on each other’s progress. “We had essentially two classes of people working on this,the people building these agents…and quantum scientists,” Anant explained. “The quantum scientists can understand the edits the agents have made, and they understand the science in ways that can help the agents incorporate changes much faster than they would on their own.”

Within 8 hours, the crowdsourcing effort matched Google’s results. In about 72 hours, it surpassed them. By the end of June, the open network had produced a circuit 47.5 percent more efficient than Google’s for overcoming 256-bit ECC. “We absolutely did not expect to beat Google,” Anant said.

Independently, at the same time Eigen Labs launched its effort, Schrottenloher published results matching Google’s, citing much of the same research. “I just put two and two together,” he said. He noted that replication was inevitable. “Cryptography and algorithms research is curiosity-driven, and the Google Quantum AI paper generated a lot of curiosity.”

Sreeram Kannan, Eigen Labs’ founder, believes the agents in his network saw Schrottenloher’s work and used it to significantly improve their results. “That’s the pace at which science can work with an open network,results built on others’ research in minutes instead of months,” he said.

Sam Jaques, an assistant professor at the University of Waterloo, described the mission to match Google’s results as a near-perfect test case for Eigen Labs’ approach. “It makes sense that AI is good at microscale optimization,” he said. “The thing about these quantum circuits is that there are a lot of places to boost efficiency here and there that may be hard for a person to see.”

A Test Case for Zero-Knowledge Proofs

The experience has cast doubt on the value of zero-knowledge proofs in research. “There is almost no situation in research where one research group is so far ahead of all the other research groups that they can keep novel results secret for long,” Jao said. “Research is an extremely competitive environment, and no team stays ahead of the curve for very long. I believe even classified research labs no longer hold any significant advantage over the research community at large.”

Craig Gidney acknowledged in a blog post that using zero-knowledge proofs moving forward is not the right strategy. “The benefits are negligible, and the costs are many. We should just publish openly.”

For Kannan, the findings represent the first major public proof of concept for Eigen Labs’ model of open agent-based science. “We want to create frameworks to help anyone innovate,” he said. “We see two pathways ahead,one where OpenAI and Anthropic use AI to do all of science, and the rest of us just consume the results, and another where we’re coordinating with agents and others to actively shape science with our ideas, skills, and expertise. The former just sounds so disastrous to me. We all want individual agency.”

Eigen Labs is already applying its agent-based open science model beyond quantum AI. “We’ve lined up scientists in very different fields, such as materials science and biology, to tackle many different problems,” Kannan said. “We see the role of scientists as architecting the right problem for a community of agents to make progress on.”

Regarding the security implications, Dustin Moody, a mathematician at the National Institute of Standards and Technology, said the need to migrate to PQC algorithms was already imperative before these results. The new findings from Google, Eigen Labs, and Schrottenloher, he noted, “seem like they are helping some people be more convinced they can’t put this off and should actually accelerate their migration plans. If an organization can migrate more quickly, it seems like a good idea to do so.”