7 min read · 11 sources

Ethereum researcher Justin Drake wants the industry to enter an operational “bunker mode”. The trigger is not quantum computing, but recent leaps in automated mathematics, including OpenAI’s release of specialized mathematical models that have begun dismantling long-standing mathematical conjectures. Drake warns that elliptic curves carry algebraic structure that machine-discovered techniques could exploit to derive private keys from exposed public keys within weeks on large GPU clusters.

The fix is deceptively simple: move capital to clean addresses that have never broadcast an outbound transaction. Ethereum addresses are derived from the Keccak-256 hash of a secp256k1 public key. Until an address initiates a signed transaction, its raw public key remains hidden behind that preimage-resistant hash. The moment an address signs anything, its public key enters the mempool and stays permanently visible onchain.

Drake is not calling for new cryptographic primitives or emergency hard forks today. Instead, he advises immediate address-rotation hygiene: hold balances strictly on virgin addresses, route multisig signatures offchain, and require oracles and L2 security councils to rotate one-time ECDSA keys per signature.

The absence of cryptographic breakthroughs among OpenAIs 722 published mathematical results raises the possibility of government-suppressed cryptanalysis.

Vitalik backs hash-based cryptography over lattices

Source: threadreaderapp.com ↗

Vitalik Buterin took Drake’s warning a step further, outlining why Ethereum should avoid NIST-standardized post-quantum schemes like ML-DSA, Falcon, and general lattice-based primitives (such as Learning With Errors and FHE). Buterin argues AI models could compress 50 years of human mathematical cryptanalysis into two years. He drew parallels to how the General Number Field Sieve systematically degraded RSA’s effective bit-security over decades of incremental algorithmic optimization.

Ethereum’s lean post-quantum roadmap is ditching lattices and lattice-based zero-knowledge commitments entirely. The target is hash-only signatures: Winternitz One-Time Signatures (WOTS) and SPHINCS+. Hash functions lack the rich algebraic structures that elliptic curves and lattice problems expose to algorithmic solvers.

The rub for protocol architects is that you cannot build public-key encryption from hash functions alone. While transaction authorization can survive on hash trees, private messaging, VPN handshakes, and onchain encrypted state still require asymmetric hardness assumptions. For now, Buterin advises moving multisig coordinator confirmations offchain and passing encrypted payloads over point-to-point channels rather than publishing raw ciphertexts to public ledgers.

The industry pushes back: panic vs. sound opsec

Source: theblock.co ↗

The security community is sharply divided over the severity of Drake’s warning, as reported by The Block. Yehuda Lindell, head of cryptography at Coinbase, dismissed the idea of a near-term ECDSA collapse as “FUD,” arguing there is zero evidence that the hardness of the discrete logarithm problem over secp256k1 has been mathematically degraded.

Dragonfly partner Haseeb Qureshi took the opposite stance, calling Drake’s warning a “very sober call” given the opaque nature of closed frontier model labs. Drake himself noted that among OpenAI’s 722 published automated mathematical proofs, the total absence of cryptographic results suggests that dual-use cryptanalysis discoveries may already be subject to voluntary withholding or government secrecy orders.

Over on Solana, Foundation engineer Jacob Creech pointed out that Solana accounts use Ed25519 rather than secp256k1, but conceded that public keys are visible by default in account models. Creech argued Solana can transition directly to post-quantum security via hash-based state proofs without forcing end-users into address-rotation fire drills.

Standard Chartered expands custody to Singapore

Source: links.tldrnewsletter.com ↗

Standard Chartered is rolling its institutional custody product into Singapore by the end of 2026, subject to approval from the Monetary Authority of Singapore (MAS). The rollout will cover major crypto assets, fiat-backed stablecoins, and tokenized real-world assets (RWAs).

This move directly integrates the bank’s internal infrastructure with Zodia Custody, the institutional custodian it originally incubated alongside Northern Trust before bringing it fully in-house. Singapore joins Standard Chartered’s active digital asset custody desks in Hong Kong, Luxembourg, and the UAE.

For institutional allocators, bank-grade custody removes the counterparty risk of holding private keys with standalone venture-backed custodians. Regulated balance-sheet custody is rapidly shifting from a value-add service into baseline compliance infrastructure across Asian financial hubs.

Polygon opens TRON liquidity through its Open Money Stack

Source: links.tldrnewsletter.com ↗

Polygon Labs expanded its Open Money Stack to support TRON, targeting the massive pool of USDT circulating on the network. The integration gives fintechs and cross-border payment processors a single SDK to route fiat bank balances directly into TRON-based USDT.

The infrastructure combines two recent Polygon acquisitions: Sequence (for smart-contract and programmable account abstraction) and Coinme (securing regulated money transmitter licenses across 48 US states). The integration also introduces cross-chain settlement logic that routes liquidity dynamically between TRON and EVM-compatible rollups. Polygon claims its underlying stack has handled over $3 trillion in aggregate transaction volume. TRON remains the dominant chain for non-US dollar remittances, making native EVM-to-TRON payment orchestration a high-margin gateway for institutional payment rails.

Consumer credit meets arbitrary collateral via Reap

Source: threadreaderapp.com ↗

Fintech issuer Reap introduced Reap Virtual Assets, an issuing engine that makes arbitrary onchain collateral spendable over existing Visa rails. Rather than requiring users to manually swap assets to fiat before swiping, the protocol allows developers to back card limits using BTC, tokenized equities, physical gold tokens, yield-bearing vaults, or raw reward points.

The system pulls price feeds directly from live oracle updates or fixed vendor rates to calculate buying capacity on the fly. Collateral remains locked in the issuing protocol’s vault contracts, while Reap executes backend settlement with Visa in USDC or USDT. It replaces complex manual liquidation loops with automated, just-in-time stablecoin clearing.

Agentic finance reports massive Q3 traction

Source: links.tldrnewsletter.com ↗

Cambrian Network published its Q3 2026 Agentic Finance Landscape Report, detailing autonomous software agents executing onchain financial operations. The metric that matters: machine-native payment protocol x402 cleared 232 million transactions for $54 million in volume, proving that AI agents operate on sub-cent, high-frequency micropayments that break traditional card fee models.

Adoption is shifting up the stack into regulated entities. ERC-8004 agent registrations doubled quarter-over-quarter to 500,000 across 22 chains. On the consumer side, Robinhood recorded over 100,000 accounts opened under its Agentic Trading product, while Base introduced Coinbase Advisor, the first SEC-registered algorithmic investment entity run directly by autonomous agents.

Concurrently, a consortium of 21 Tier-1 banks, including Citi, Goldman Sachs, and Bank of America, announced a dedicated stablecoin joint venture to handle interbank agentic settlement. Machine-to-machine commerce is actively forcing payments giants to adapt: Mastercard launched Agent Pay for Machines with Stripe and Coinbase, while Visa endorsed x402 and MPP architectures for autonomous micro-transactions.

Morpho's fee mechanics diverge from Uniswap's

Source: threadreaderapp.com ↗

Discussion around Morpho activating its protocol fee switch misses key differences in lending protocol mechanics, according to analysis contrasting Morpho’s model with Uniswap’s historical fees.

In automated market makers, fee capture is offset by toxic order flow protection and concentrated volume that sticky LPs can harvest. In Morpho’s isolated lending markets, activating a fee switch directly depresses lending yields or compresses vault curator margins. Because lending capital is entirely portable, imposing rent on curators risks migrating deposits to Aave V4. Furthermore, Uniswap’s recent revenue bump was driven by short-lived memecoin volume on Robinhood, with platform trading volume subsequently falling 71% from its September high. Morpho will need to prove actual supplier pricing power before it can skim protocol revenue without shedding total value locked.

Quick hits: PropAMMs, macro drawdowns, and BitMine’s ETH ceiling

Source: ethresear.ch ↗

• PropAMMs struggle for market share on Ethereum: Proprietary AMMs, which combine offchain pricing with onchain settlement, represent 5% to 10% of Ethereum swap volume, compared to roughly 40% on Solana. Between September 14 and September 20, Fermi processed $280 million and Metric settled $178 million, remaining far behind Uniswap’s $6 billion weekly run rate. On Solana, BisonFi processed $2.8 billion alone, highlighting the structural latency advantages high-throughput L1s hold for private-market-maker DEX designs.
• Geopolitical shocks liquidate $550M in crypto longs: Bitcoin slipped 1.6% to touch $82,800, blowing past technical support around $83,000 and putting an $80,000 retest on the table. Brent crude climbed above $102 a barrel on reports of US strike planning against Iran, pushing the US 10-year Treasury yield up to 5.31%. Over $550 million in leveraged long derivatives were wiped out across major exchanges within twelve hours.
• BitMine caps its Ethereum accumulation at 5%: Treasury vehicle BitMine disclosed it held 6,016,414 ETH as of October 4, mirroring MicroStrategy’s corporate treasury blueprint. Executive Tom Lee stated the company will permanently stop buying ETH once it reaches a hard limit of 5% of circulating supply. BitMine requires roughly 100,000 ETH to hit that cap, removing a persistent programmatic spot buyer from the order books in the near future.