AI is forcing crypto to rethink where it keeps its valuables, DeFi rallies to cover losses, and a proposed Bitcoin fork is stirring a fresh property-rights fight — here’s what’s happening.
Mythos shifts the threat model from contracts to infrastructure
Anthropic’s new adversary-style AI, Mythos, isn’t just unsettling banks and big tech — it’s changing how crypto teams prioritize security. For years, DeFi’s hardening efforts centered on smart contracts: audits, bug bounties and cataloged exploit patterns. Mythos flips that focus toward the less-visible systems that glue protocols together.
“My concern with AI-driven threats is less about smart contract exploits and more about AI-assisted attacks against the human and infrastructure layers,” said Paul Vijender, head of security at risk firm Gauntlet. Mythos is built to simulate attackers across whole systems, not merely scan for known code bugs. It looks for ways small weaknesses can be chained together to become real-world exploits — particularly targeting key management, signing services, bridges, oracle networks and the cryptographic plumbing that connects them.
Those components often sit outside traditional audits and are harder to observe. The point was driven home when web-hosting provider Vercel disclosed a breach this month that may have exposed customer API keys, forcing crypto teams to rotate credentials and re-check deployments. Vercel traced the incident to a compromised Google Workspace connection used via the third-party AI tool Context.ai by an employee.
The risk is being taken seriously across industries. Major banks are treating AI-driven cyber risk as systemic and exploring similar stress-testing tools, and reports say Coinbase and Binance approached Anthropic about Mythos testing. Early results from such models are flagging vulnerabilities in the operational layers that keep platforms secure — the same layers Mythos is designed to stress.
Aave coordinates a rare industry-wide recovery: $301M pledges
When a Kelp DAO exploit destabilized rsETH markets and put lending positions at risk on Aave, the response was unusually fast and coordinated. An effort dubbed “DeFi United” had amassed about $301 million in commitments, according to its website, much of it awaiting governance approval.
Aave’s DAO is central to the plan: a governance proposal contemplates allocating up to 250,000 ETH toward recovery, and founder Stani Kulechov pledged 5,000 ETH personally. Individuals and firms around Aave have stepped up as well — contributions and pledges include Emilio Frangella (500 ETH), BGD Labs’ Ernesto Boado (100 ETH) and BGD Labs itself (250 ETH), plus Marcelo Ruiz de Orlano (100 ETH).
The outreach widened quickly. After a related April 18 bridge hack, Kulechov reportedly contacted ConsenSys and other ecosystem players to coordinate aid. ConsenSys and founder Joseph Lubin agreed to commit up to 30,000 ETH in support, with advisory input from Sharplink in planning stages.
“This is about supporting users and restoring normal market conditions,” an Aave Labs spokesperson said. The episode underscores a maturing DeFi ecosystem where capital, infrastructure providers and governance actors can mobilize together to limit contagion.
“Crypto was built for AI agents,” says Alchemy CEO — and that’s an advantage
Nikil Viswanathan, co-founder and CEO of infrastructure provider Alchemy, argues crypto’s architecture aligns naturally with the needs of autonomous AI agents. Traditional finance, he told CoinDesk, was designed for human constraints — geography, time zones, paperwork and physical identities. Agents don’t sleep, don’t occupy a single country and don’t carry cards; they transact online, globally, and often in tiny, programmable increments.
“Crypto is the global infrastructure for money that agents need,” Viswanathan said, adding that agents require always-on, programmable, borderless rails with direct control via code. He’ll be speaking about these themes at Consensus Miami next month. The broader point: as AI agents begin to act as economic participants, crypto’s composable, permissionless infrastructure may shift from niche alternative to foundational layer for machine-driven commerce.
eCash fork revives debate over “copying” Satoshi’s coins
A proposed Bitcoin fork called eCash — scheduled to launch in August at block height 964,000 — is reigniting questions about what a chain copy really means for ownership. The fork would mirror Bitcoin’s ledger up to the split, so BTC holders would receive an equal balance on the new chain (for example, 4.19 BTC => 4.19 eCash), following the model used by Bitcoin Cash and Bitcoin SV.
The controversy centers on roughly 1.1 million BTC attributed to Bitcoin’s pseudonymous creator, Satoshi Nakamoto. On a straightforward fork those addresses would receive about 1.1 million eCash. Paul Sztorc, CEO of LayerTwo Labs and eCash’s lead, says his plan would credit 600,000 eCash to those dormant addresses and allocate the remaining 500,000 eCash to investors who fund the project before launch.
Sztorc pushed back on characterizations of theft in a post on X, noting that eCash cannot move BTC on the original chain because it lacks Bitcoin private keys or software. Yet critics argue that monetizing claims on a forked-chain version of Satoshi’s holdings — even if the original BTC remain untouched — undermines the symbolic property right that Satoshi’s unmoved coins represent for Bitcoin: that the creator never spent them under the same rules everyone else follows. That dispute has turned into a broader property-rights debate over what it means to “own” assets that exist on a fork.
What to watch next
- How Mythos-style AI testing reshapes audits and third-party risk assessments across crypto infrastructure.
- Governance votes at Aave and whether pledged commitments translate to on-chain support.
- Institutional interest in agent-native finance as AI adoption grows.
- Community and market response to eCash as the fork date approaches.
Reporting by Margaux Nijkerk, Will Canny and Shaurya Malwa.
Read more AI-generated news on: undefined/news
