CZ
Oct 23, 2026

CZ says AI can amplify phishing and social engineering risks without breaking Bitcoin cryptography

Changpeng Zhao, widely known as CZ, said concerns about AI attacking crypto systems are common, but noted that internet and banking systems in practice rely on weaker encryption technologies. He argued that AI does not need to actually break Bitcoin’s cryptographic design to cause damage. According to CZ, existing data leaks are already enough to let AI scale harmful activity in areas such as phishing and social engineering. His remarks shift the focus away from a direct compromise of Bitcoin’s core cryptography and toward the more immediate security risks created when leaked user data is combined with AI tools. The comments highlight a threat model centered on abuse of exposed information rather than a technical break of Bitcoin itself.

670
CZ says AI can amplify phishing and social engineering risks without breaking Bitcoin cryptography
AI
Oct 20, 2026

AI-driven cryptography concerns push Web3 to prepare early, but not rush wallet migrations

A fresh debate over whether AI could accelerate breakthroughs in cryptanalysis has stirred concern across the crypto industry, but the core message from the discussion is not that users should immediately move funds. In an article published by imToken and carried by Foresight, Ethereum researcher Justin Drake called on the blockchain sector to begin planning a "Bunker Mode" response to future cryptographic risks, warning that advances in AI-assisted mathematics could shorten the timeline for new attacks on elliptic curve cryptography. Vitalik Buterin joined the discussion with a different emphasis: while he did not dismiss the possibility that AI could speed up cryptographic research, he argued that rushed upgrades and asset migrations may create more immediate harm than attacks that have not materialized. The article explains that the issue turns on how public keys are exposed, why addresses and public keys are not always the same thing, and why moving assets to a fresh address is only a temporary risk-reduction step rather than a full post-quantum migration. It also reviews how Ethereum, Bitcoin and NEAR are approaching post-quantum preparation, and argues that real readiness will require protocol changes, wallet support, ecosystem coordination and careful user-side execution rather than one-click fixes.

730
AI-driven cryptography concerns push Web3 to prepare early, but not rush wallet migrations
Anthropic
Oct 19, 2026

Claude reconstructs NSA seed origins for NIST curves, easing a long-running backdoor suspicion

Anthropic says Claude reconstructed how the U.S. National Security Agency generated the public seeds used for a set of NIST elliptic curves in the late 1990s, solving a cryptographic history puzzle that had lingered for decades. The result does not mean AI broke elliptic-curve cryptography, derived anyone’s private key, or compromised TLS or Bitcoin. Instead, it explains how the published seeds behind core NIST prime curves appear to have been produced. According to the report, Claude examined 34 legacy pseudorandom number generators and output-slicing patterns, identified traces consistent with the POSIX/C library function lrand48(), and used LLL lattice reduction to recover internal state from truncated outputs. Anthropic said the model reached the method in under 30 minutes. The reconstructed process points to specific UTC timestamps between 1997 and 1998 and candidate positions that reproduce the published 160-bit seeds exactly. The finding matters because the seeds had long been auditable only in one direction: anyone could verify that a given seed generated a given curve coefficient, but not why those 160-bit seeds were chosen in the first place. Filippo Valsorda, who helped set a bounty on the problem in 2023, independently reimplemented the process in Go and reproduced the seeds, giving the result an external check. The report argues that this sharply narrows one concrete class of curve-selection backdoor theories while leaving some parameter-selection details unresolved.

1230
Claude reconstructs NSA seed origins for NIST curves, easing a long-running backdoor suspicion
a16z
Oct 18, 2026

a16z Says AI Has Not Broken Modern Cryptography, but Could Speed Up Security Reviews

a16z crypto said recent advances in AI-driven mathematics have not shown that modern cryptographic systems are suddenly weaker, pushing back on fears that new models are on the verge of breaking the foundations of digital security. In a discussion featuring Stanford cryptographer Dan Boneh, a16z crypto general partner Eddy Lazzarin, research partner Justin Thaler, investment partner Liz Harkavy, and editor Sonal Chokshi, the group argued that the latest results belong to mathematics and computer science, not cryptanalysis. The speakers said there is still no evidence that elliptic-curve cryptography has become more fragile or that any specific cryptographic assumption is less secure than before. Thaler added that much of modern cryptography rests on the belief that certain efficient algorithms do not exist at all, meaning a more capable AI system would not automatically find one if none exists. Boneh also pointed to the unresolved P versus NP problem as a reminder that cryptography still depends on assumptions that remain unproven. Where AI may matter sooner, they said, is in compressing years of security review into much shorter cycles, helping researchers and companies test assumptions, audit software, and uncover vulnerabilities before deployment. The discussion also noted that work already done for quantum-readiness and cryptographic agility could help systems respond faster if new weaknesses emerge.

1040
a16z Says AI Has Not Broken Modern Cryptography, but Could Speed Up Security Reviews
Anthropic
Oct 17, 2026

Claude traces NIST curve seeds in under 30 minutes, easing NSA backdoor concerns

Anthropic said on Oct. 16 that its AI model Claude identified the origin of the seeds used in NIST standard elliptic curves in less than 30 minutes, offering a new explanation for a long-running cryptographic mystery. Those seeds were supplied by the U.S. National Security Agency, or NSA, when NIST published the curves in the FIPS 186-2 digital signature standard in 2000, but their source was never documented, leaving room for suspicions that the parameters may have been chosen to hide a backdoor. According to Anthropic, Claude found that the seeds were not hashes of secret phrases, as some had speculated, but outputs from the C standard library random number generator lrand48 initialized with Unix timestamps. The company said the finding is not a cryptographic break, a new attack, or a weakness in ECDSA curves. Instead, it argues that the result makes deliberate manipulation less likely because the process appears reproducible. The research covered five NIST prime curves and five ANSI X9.62 example curves, with matching timestamps between May 1997 and April 1998. Some questions remain unresolved, including the origins of several base points and the seed for the binary curve B-163. Anthropic also said Bitcoin’s secp256k1 curve was not part of the study.

1140
Claude traces NIST curve seeds in under 30 minutes, easing NSA backdoor concerns
Anthropic
Oct 17, 2026

Claude traces disputed NIST curve seeds to old RNG in under 30 minutes

Anthropic said its AI model Claude solved a 25-year cryptography mystery in less than 30 minutes by tracing a set of numbers supplied by the U.S. National Security Agency in the late 1990s to an ordinary random number generator that used 1997 and 1998 timestamps. The numbers had long drawn suspicion because the National Security Agency never explained where some of the seeds used to generate elliptic curves later adopted by the U.S. National Institute of Standards and Technology in 2000 came from. Those concerns intensified after Edward Snowden’s disclosures that the NSA had tried to weaken cryptographic standards. The issue had been discussed by cryptographers Neal Koblitz and Alfred Menezes in a 2015 paper, pursued for years by security researcher Steve Weis, and turned into a bounty challenge in 2023 by Filippo Valsorda and others. Anthropic said Claude traced five NIST curve seeds to the lrand48 random number generator in an older C software library and reproduced the widely used P-256 curve sequence, finding an exact match for the published seed at candidate No. 1842. The company said the result sharply reduces the likelihood of a hidden weakness and that users of NIST curves do not need to make changes. Bitcoin is not affected because its wallets use the simpler secp256k1 curve. The $36,864 bounty will be donated to Internet Security Research Group, the nonprofit behind Let’s Encrypt.

1160
Claude traces disputed NIST curve seeds to old RNG in under 30 minutes
Claude
Oct 17, 2026

Claude resolves long-running NSA elliptic curve seed mystery in under 30 minutes

Anthropic said its AI model Claude solved a cryptography puzzle that had lingered for about 25 years, tracing the origin of elliptic curve seeds adopted by the National Institute of Standards and Technology, or NIST, in 2000. According to the report, Claude linked five NIST elliptic curve seeds to the lrand48 random number generator in an older C software library used in 1997 and 1998. It then found a value that exactly matched the widely used P-256 curve seed at candidate No. 1842, allowing the generation process to be reproduced. The finding addresses a long-standing concern around whether the National Security Agency, or NSA, may have embedded weaknesses in those parameters, since the agency had never fully explained their origin. Anthropic said the result sharply reduces the likelihood that the parameters were deliberately backdoored and that current users of NIST elliptic curves do not need to take extra action. The company also noted that Bitcoin is unaffected because it uses a different elliptic curve, secp256k1. A $36,864 research bounty tied to the work will be donated to Internet Security Research Group, the nonprofit organization behind Let's Encrypt.

960
Claude resolves long-running NSA elliptic curve seed mystery in under 30 minutes
Anthropic
Oct 17, 2026

Claude cracks a 25-year NIST curve seed mystery in under 30 minutes, reviving questions around Bitcoin’s curve choice

Anthropic said on Oct. 17 that its internal model Claude reconstructed the long-unclear seed generation method behind parts of the NIST standard elliptic curves in less than 30 minutes, addressing a question that has lingered in cryptography circles for roughly 25 years. According to the disclosure, the model’s reverse-engineering work suggests the seeds were likely produced with the lrand48 pseudorandom number generator from a 1990s-era C library combined with Unix timestamps, rather than by a complex secret algorithm. The finding matters for crypto because ECDSA underpins digital signatures used across many wallets and identity verification systems. Around 2000, some parameters for NIST standard curves were supplied by the U.S. National Security Agency, while the seed generation process remained opaque, fueling longstanding speculation about possible backdoors. The report also draws a contrast with Bitcoin, where Satoshi Nakamoto chose the secp256k1 curve instead of the NIST curves commonly used elsewhere. Anthropic’s disclosure does not say ECDSA has been broken, and it does not claim any new mathematical weakness has been found in the related curves. Reconstructing how some parameters were generated may help reassess backdoor concerns, but it does not prove that every implementation is risk-free. The episode also points to a broader use for AI in security work: auditing historical cryptographic design choices and testing old assumptions.

1000
Claude cracks a 25-year NIST curve seed mystery in under 30 minutes, reviving questions around Bitcoin’s curve choice
Total 127 items
Go topage