A seed-generation flaw in Coldcard hardware wallets has been linked to an ongoing theft campaign that has affected more than 4,500 addresses and resulted in the loss of nearly $90 million in Bitcoin as of Sunday, according to comments cited by Odaily from Kraken Chief Security Officer Nick Percoco. Percoco said the issue exposed a gap in independent testing for hardware wallets. Auditors verified that the intended random number generator existed, but did not confirm that the production firmware actually called that generator. He said the missing step left room for a critical failure in how wallet seeds were created. Coinkite said the software flaw had existed since March 2021. During the integration of a new cryptographic library, the wallet-creation flow was mistakenly routed to a weaker MicroPython generator. Percoco added that the hardware-wallet sector lacks end-to-end validation procedures comparable to NIST SP 800-90B and BSI AIS-31, and that current certifications and vendor-commissioned audits do not systematically require proof that production firmware uses a validated entropy source. Coldcard said it halted all device shipments after confirming the flaw on Thursday and destroyed all remaining devices at its facility that contained the affected firmware. Coinkite also told impacted users not to discard their devices and said its legal team may coordinate with law enforcement across multiple jurisdictions.

WuBlockchain’s July technology report mapped a broad set of protocol developments across Bitcoin, Ethereum, Solana, BNB Chain, Zcash, NEAR, Cardano, Uniswap and Sui. On Bitcoin, Bitcoin Core v31.1 was released on July 8, 2026 with bug fixes tied to chainstate disk rewrites and an IP leak linked to the -privatebroadcast feature, while BIP-110 turned into a governance flashpoint as miner signaling stayed far below the 55% threshold. On Ethereum, the Glamsterdam upgrade moved into its final integration and performance-testing phase, Hegota remained in proposal-scope selection, and Vitalik Buterin published new pieces on Lean Ethereum, validator-state compression and privacy. The report also covered Ethereum proposals including Native UTXOs on Ethereum and EIP-8222, along with Starknet’s STRK20 privacy framework and Polygon’s Ithaca hard fork schedule. Elsewhere, Solana advanced Agave v4.2, launched Solana Governance Proposals and prepared SIMD-0286 to raise the single-block compute cap, while Zcash activated Ironwood NU6.3 on mainnet and introduced Zakura 1.0.0 as a new full-node client.

Anthropic’s Frontier Red Team said its internal model, Claude Mythos Preview, identified two cryptographic weaknesses in largely autonomous research runs, but neither result breaks encryption used in practice today. One finding targeted HAWK, a post-quantum digital signature candidate under review at the U.S. National Institute of Standards and Technology, where the model found a previously unknown symmetry in the lattice structure and cut the attack cost for HAWK-256 from 2^64 to 2^38. The work took about 60 hours and roughly $100,000 in API usage. The second result focused on a reduced 7-round version of AES-128 rather than the full 10-round AES. Anthropic said the model developed a fingerprinting method called Möbius Bridge that removed one enumeration step and ran 200 to 800 times faster than earlier meet-in-the-middle attacks. That effort took about three days and billions of tokens, with humans providing only limited guidance. Anthropic framed the results as evidence that AI can conduct cryptanalysis research, not as proof that Claude has broken modern deployed encryption. The targets were an unadopted candidate scheme and a deliberately weakened AES variant, leaving real-world systems unaffected.

Anthropic said an unreleased version of its flagship model, Claude Mythos Preview, discovered two previously unknown attacks on cryptographic algorithms, including one targeting HAWK, a post-quantum digital signature system that had advanced to the third round of the U.S. NIST competition. According to the report, the attack cuts the cost of stealing HAWK’s smallest key from 2^64 operations to 2^38, or about 67 million times less work. Anthropic said fixing the issue would roughly double key sizes, weakening one of HAWK’s main selling points: compact keys and fast signing. The report also stressed that HAWK has never been deployed and that Bitcoin still uses ECDSA, so no live crypto system was broken by the finding. Even so, the development matters because a broad push toward quantum-resistant crypto infrastructure assumes that the replacement schemes are themselves secure. The newsletter also covered a range of market and policy developments, including gains in BTC and ETH ahead of the FOMC, public backing for the CLARITY Act from BlackRock, Fidelity, Franklin Templeton, Goldman Sachs, and SoFi, fresh ETF flow data, Zcash’s Ironwood activation, and a burst of meme coin trading activity led by PIPEDOG on Robinhood chain.

Anthropic said an unreleased version of its Claude Mythos model uncovered two previously unknown cryptographic attacks, including one against HAWK, a post-quantum digital signature scheme now in the third round of the U.S. National Institute of Standards and Technology’s competition. According to the company, Claude identified a symmetry in HAWK’s structure that cuts the cost of recovering a secret key in the smallest configuration from 2^64 operations to 2^38, a roughly 67 million-fold reduction. Anthropic said fixing the issue would require roughly doubling HAWK’s key size, which would undercut the compactness and speed that make the scheme attractive for blockchains and other systems sensitive to signature size. The company also described a separate result on a 7-round research version of AES-128, where Claude was asked to avoid all five established families of AES cryptanalysis and invent a new approach. Anthropic said the model eventually produced a technique called “Möbius Bridge,” reducing one of the nine key bytes an attacker previously needed to guess. It also reported that Claude broke 13 rounds of LEA, though deployed LEA uses 24 rounds. Anthropic said it disclosed the findings to the algorithms’ authors and relevant U.S. government and industry partners before publication, and coordinated the HAWK finding with NIST.

VISReg, a new self-supervised learning method built around Variance-Invariance-Sketching Regularization, has drawn repeated reposts from Turing Award winner Yann LeCun, who described the line of work as: "VICReg begat SIGReg which begat VISReg." The paper positions itself as a response to one of the central problems in JEPA-style world models: representation collapse, where distinct inputs are mapped to the same or only a few vectors and the model stops learning useful features. The method splits anti-collapse regularization into two independent targets, scale and shape. It keeps a variance term to control scale, replaces covariance constraints with a sketching objective based on Sliced Wasserstein Distance to capture distribution shape, and uses stop-gradient to separate the two during optimization. According to the paper, this avoids the vanishing-gradient issue seen in SIGReg when collapse begins. The authors report results across 15 datasets, including 8 in-domain benchmarks, 6 out-of-distribution datasets, and ADE20K dense prediction. The paper says VISReg outperformed seven mainstream self-supervised methods in aggregate and matched DINOv2 on OOD benchmarks using about one-tenth of the training data. Code, pretrained weights, and the paper are publicly available.

Ethereum’s long-range research agenda is being consolidated into a single framework that stretches from the planned Glamsterdam upgrade in the second half of 2026 through a series of hard forks expected to run into 2029. The draft document, known as “Strawmap,” follows Justin Drake’s Lean Ethereum vision and groups the network’s ambitions into five “North Star” goals. Those goals cover a faster Layer 1 with finality cut from roughly 15 minutes to seconds, a jump in Layer 1 throughput from about 5 million gas per second to 1 billion gas per second, a Layer 2 data path that could scale to 1 GB per second, a post-quantum cryptography transition, and native privacy features at the base layer. The proposal remains a draft and is still being revised. Glamsterdam has already slipped from the first half of 2026 to the second half. Even so, the document offers one of the clearest looks yet at how Ethereum researchers see the chain evolving by 2030: a settlement-focused base layer, a larger role for ZK proofs across consensus and execution, a wider field of independent development groups alongside a smaller Ethereum Foundation, and a network designed to be faster, more trust-minimized and more resilient to future cryptographic threats.

Algorand Foundation has outlined a post-quantum roadmap starting in 2026, with post-quantum accounts, multisig wallets and staking support, aiming for broad quantum resilience across the network by the end of 2027.