Anthropic
Oct 16, 2026

Anthropic Says Claude Linked NIST Curve Seeds to 1990s Timestamps

Anthropic said Claude traced NIST curve seeds back to timestamps from the 1990s, according to a brief report by The Defiant. The report stressed that the result does not amount to a cryptographic break, drawing a line between historical attribution and any claim of weakness in the curves themselves. It also said published code covers five NIST prime-curve seeds and four binary-curve seeds. The article additionally noted that its content is stored on Filecoin. While the source text was brief, the core points were clear: Anthropic is framing the work as seed tracing rather than a break, and the scope disclosed in the published code spans nine curve seeds in total across prime and binary categories.

1030
Anthropic Says Claude Linked NIST Curve Seeds to 1990s Timestamps
Ethereum
Oct 16, 2026

Can ETH hit another record? Foresight article says the answer depends on the investor

A market analysis piece published by Foresight tackled two questions at once: whether Ether can set a new all-time high, and how seriously the industry should take the idea that AI could break existing cryptographic systems. On ETH, the article did not present a fresh price target or trading call. Instead, the author said the core issue comes down to each investor’s own conviction, arguing that outside opinions can only serve as reference points. To illustrate that point, the author mentioned often citing Duan Yongping’s methodology while still feeling unable to buy Apple, his largest U.S. stock holding, because conviction matters more than imitation. On AI and cryptography, the article said recent discussion has been driven by two factors: OpenAI publishing more than 700 papers that reportedly claim to solve long-standing mathematical problems, and concern from a senior Ethereum core researcher that AI’s rapid progress could threaten elliptic-curve-based cryptography even before quantum computing reaches large-scale commercial use. The piece also summarized Vitalik Buterin’s idea that AI might discover hidden mathematical patterns or vulnerabilities in current cryptographic systems. Still, the author argued that both the quantum-computing route and the AI-discovery route require a full public key to be exposed, and said moving assets to newly generated addresses that have never signed outgoing transactions is currently a practical way to reduce that risk.

630
Can ETH hit another record? Foresight article says the answer depends on the investor
Google
Oct 16, 2026

Google says Willow quantum chip cut magic-state error rates by about 40x in lab test

Google Quantum AI said on Oct. 15 U.S. time that it had experimentally demonstrated "magic state cultivation" on its Willow quantum chip for the first time, a result the company described as a step toward easing one of quantum computing’s most expensive bottlenecks. According to Google’s official blog, the related paper was published the same day in Science. In the experiment, Google said the cultivated magic states showed error rates about 40x lower than uncultivated ones, while the logical fidelity of the T state reached 99.99%. The team also demonstrated moving the state from a distance-3 color code to a distance-5 surface code. Still, Google said practical use will require larger-scale demonstrations and that major technical challenges remain. Decrypt reported that the experiment used at most 54 of Willow’s 105 qubits, with roughly 8% of attempts passing all checks. It also said the result is still about one million times away from the cleanliness needed to factor RSA-2048, meaning crypto wallets do not face immediate risk from this work.

850
Google says Willow quantum chip cut magic-state error rates by about 40x in lab test
Google
Oct 16, 2026

Google says Willow cut magic-state errors by 40x, reaching 99.99% fidelity

Google Quantum AI said its Willow processor has demonstrated a technique called magic state cultivation, reducing the error rate of magic states by about 40x and producing T states with 99.99% logical fidelity, according to a paper published in Science. The result targets one of quantum computing’s most stubborn cost barriers: generating the high-quality magic states required for complex, commercially useful quantum programs. Google said prior academic estimates suggested magic-state production could consume as much as 95% of a quantum computer’s total resources. By shifting from the older “magic state distillation” approach to a more selective cultivation process, the company argues that the physical overhead for useful quantum machines could fall sharply. In the experiment, only about 8% of attempts passed all checks, meaning more than 90% of initial candidates were discarded to keep only the cleanest states. The work also feeds into long-term estimates around cryptography. The report cited earlier projections from researcher Craig Gidney that breaking RSA-2048 may require fewer than 1 million noisy physical qubits, down from a 2019 estimate of 20 million. A separate white paper from Google Quantum AI, Ethereum Foundation researcher Justin Drake, and Stanford cryptographer Dan Boneh estimated that breaking Bitcoin’s elliptic curve cryptography could require fewer than 500,000 physical qubits. Even so, the paper says the error rate needed to factor RSA-2048 remains far below what Willow has reached, and no crypto wallet is at immediate risk.

860
Google says Willow cut magic-state errors by 40x, reaching 99.99% fidelity
Google
Oct 15, 2026

Google says Willow chip test cut quantum magic-state errors by about 40x

Google says it has moved one of quantum computing’s most discussed ideas out of theory and onto working hardware. In a paper published Thursday in Science, the company’s Quantum AI team reported that a technique called magic state cultivation reduced errors by about 40 times on its Willow processor and produced T-states with a record logical fidelity of 99.99%. The result matters because fault-tolerant quantum computing depends not only on standard error correction, but also on the ability to generate high-quality “magic states” for harder operations that cannot be protected the same way. Google said producing those states has historically been so expensive that some estimates put the cost at as much as 95% of a machine’s total capacity. The new cultivation approach aims to lower that burden by filtering and growing usable states inside a small chip region instead of relying on large qubit factories. The announcement also feeds into the long-running discussion around quantum threats to modern cryptography. The article links cultivation to Craig Gidney’s 2025 estimate that RSA-2048 could be broken with fewer than 1 million noisy qubits, down from 20 million in 2019, and cites a March Quantum AI whitepaper co-authored with Ethereum Foundation researcher Justin Drake and Stanford cryptographer Dan Boneh that put the cost of breaking Bitcoin’s elliptic curve cryptography at under 500,000 physical qubits. Even so, Google said nobody is draining wallets with this experiment: the test used at most 54 of Willow’s 105 qubits, and factoring RSA-2048 would still require magic states roughly 1 million times cleaner than what the company has now demonstrated.

1010
Google says Willow chip test cut quantum magic-state errors by about 40x
Google
Oct 15, 2026

Google says Willow chip advance cuts quantum error rate by about 40x

Google Quantum AI said in a paper published in Science that its "magic state cultivation" technique reduced the error rate by about 40 times on the company’s Willow processor, while reaching 99.99% fidelity for the resulting magic states. Magic states are single-use resources needed to run practical quantum programs, and earlier estimates had suggested that preparing them could account for as much as 95% of a quantum computer’s total compute capacity. Google said the work turns one of quantum computing’s most promising theoretical methods into something that can be used on chip in practice. According to the Techub summary, citing Decrypt, the advance is also one of the key pieces behind Google’s 2025 projection that the number of noisy qubits required to break RSA-2048 encryption could fall below one million. The report added that lower construction costs for quantum computers could speed up applications in drug design and new materials research, while also carrying implications for the cryptography used to secure bank logins and Bitcoin wallets.

930
Google says Willow chip advance cuts quantum error rate by about 40x
Microsoft Research
Oct 15, 2026

Microsoft Research unveils Magnetar, a zero-knowledge proof system for Rust, Go and C programs

Microsoft Research has introduced Magnetar, a zero-knowledge proof system built to support proof-based general-purpose computation. The system can generate proofs that programs written in Rust, Go, or C were executed correctly, according to the announcement cited by Techub. Microsoft Research said Magnetar delivers a major speed gain over earlier systems in the same category. Proof generation is, on average, 10x to 39x faster than the previous fastest comparable systems. The research group also said Magnetar’s security relies only on standard hash functions. In its description of the project, Microsoft Research framed Magnetar as an attempt to tackle three hurdles that have limited broader zero-knowledge proof adoption: proof generation speed, usability for ordinary programmers, and a cryptographic foundation that can remain trustworthy over the long term. The release positions Magnetar as a system meant to make proving the correct execution of conventional software more practical without requiring developers to abandon mainstream programming languages.

890
Microsoft Research unveils Magnetar, a zero-knowledge proof system for Rust, Go and C programs
Bitcoin
Oct 14, 2026

Bitcoin public key exposure is emerging as a major crypto security issue for 2027

A debate over Bitcoin’s exposed public keys is moving from theory to operational planning. Ethereum core researcher Justin Drake has urged the industry to prepare for a “bunker mode” scenario by moving funds to fresh addresses whose public keys have not yet been revealed on-chain, arguing that artificial intelligence could produce mathematical breakthroughs that make practical attacks on ECDSA possible before quantum computing becomes the main threat. In his worst-case framing, holders might have only months to react. The warning has drawn both support and criticism. Vitalik Buterin said the concern should not be limited to elliptic-curve cryptography and could also extend to lattice-based systems such as ML-DSA if AI changes how hard underlying math problems really are. NEAR founder Illia said AI is challenging assumptions about mathematical complexity, but he has not seen evidence that the timeline for practical elliptic-curve attacks has materially changed. Cryptography researcher Sean Bowe also questioned the gap between the urgency of the warning and the amount of new evidence presented. On-chain data adds weight to the discussion. Glassnode researcher Rafael said about 6.26 million BTC, or 31.2% of issued supply, already have publicly visible keys on-chain, up from 24.8% in early 2021. The figures include coins exposed through address reuse, script design, and exchange-held balances. The broader question now is not only whether an attack becomes feasible, but whether the industry can migrate wallets, custody systems, and signing methods in time.

940
Bitcoin public key exposure is emerging as a major crypto security issue for 2027
Total 127 items
Go topage