News linked to both this project and an event.
Cos, founder of SlowMist, shared a tweet on X platform regarding potential poisoning attack risks in Claude Code and published a detailed analysis of poisoning attacks targeting Grok Build CLI and Claude Code CLI. The analysis pointed out that the security mechanisms of Grok Build CLI are not unified, with different code paths having different trust assumptions, and the gaps between them serve as channels for attackers.Attackers may exploit malicious project configuration files to execute arbitrary commands without the user's knowledge, thereby stealing API keys, cloud credentials, or gaining control over local devices. Researchers constructed a test environment and found that on Mac systems, if Claude Code is compromised, executing a specific test command could trigger the launch of a local calculator, demonstrating a potential command execution risk. If the attack succeeds, attackers could further steal API keys from AI services such as Claude and OpenAI, causing account cost losses; obtain credentials for cloud services like AWS, Alibaba Cloud, and Tencent Cloud to access servers and data; tamper with code repositories to implant backdoors; and leverage local devices as a springboard to attack internal enterprise networks. It is reported that the relevant vulnerability has existed for one year.
SlowMist founder Cosine retweeted a post about the potential poisoning attack risks of Claude Code, pointing out that attackers could execute arbitrary commands without the user's knowledge through malicious project configuration files, thereby stealing API keys, cloud credentials, or controlling local devices. Researchers constructed a test environment and found that in Mac systems, if Claude Code is affected, executing specific test commands can trigger the local calculator to launch, proving the existence of potential command execution risks. If the attack succeeds, attackers may further steal API Keys for AI services such as Claude and OpenAI, causing account fee losses; obtain cloud service credentials for AWS, Alibaba Cloud, Tencent Cloud, etc., to access servers and data; tamper with code repositories to implant backdoors; and use local devices as a springboard to attack enterprise internal networks.
Odaily Odaily A new study by the Cambridge Centre for Alternative Finance reveals that approximately 31% of Ethereum node activity is located in the United States, with another 39% distributed across EU countries excluding the UK, indicating that the geographic distribution of Ethereum nodes remains relatively concentrated in Western nations.Lead researcher Alexander Neumuller stated that while node distribution is not currently concentrated in any single country, it is heavily reliant on a few major cloud service providers, including Hetzner, Amazon AWS, and OVH. Notably, the Ethereum network does not require half of its validators to fail for problems to arise. If more than one-third of validators go offline simultaneously, the network may be unable to finalize block checkpoints (finalization). Neumuller pointed out that nodes and validators do not have a one-to-one correspondence; a single node may run multiple validators. Therefore, it is currently impossible to precisely assess the actual impact on the validator network from the failure of a specific node or service provider.Furthermore, the study reassessed the energy consumption of Ethereum following The Merge. Data shows that Ethereum's current annual energy consumption is approximately 7.9 GWh, equivalent to a continuous power draw of about 1 MW. This represents only about 0.02% of pre-merge levels, a reduction of approximately 99.98%. Currently, over 56% of the energy used by the Ethereum network comes from sustainable sources, exceeding the global average.The study also noted that client software diversity is another potential risk. If a dominant client software has a vulnerability, it could affect a large number of network participants. The report was published by the Cambridge Centre for Alternative Finance and supported by the Ethereum Foundation. (The)
the U.S. government's export controls and access restrictions on Anthropic's models, Fable 5 / Mythos 5, were partly driven by Amazon's cybersecurity research and AWS CEO Andy Jassy's communications with the White House.It is understood that research submitted by Amazon indicated that through a series of prompt tests, researchers could induce Fable 5 to output sensitive information potentially usable for cyberattacks, raising security concerns. Subsequently, Andy Jassy reported these findings to the U.S. government level, prompting the White House to implement further restrictions, including banning foreign users from accessing the model.Meanwhile, former U.S. Commerce Department official Kate Koren revealed that the White House's existing policy stance towards Anthropic may have also influenced this decision. This is because Anthropic has disagreements with the White House over the boundaries of AI safety, including refusing to use its models for mass surveillance or lethal autonomous weapons systems. Although the two sides had eased tensions and expanded cooperation earlier this year, this incident could reignite strained relations between them. (The Wall Street Journal)
According to on-chain analyst PeckShield (@PeckShieldAlert), SlowMist’s threat intelligence system MistEye has detected a cross-registry supply chain attack targeting developers. Malicious packages have spread across three major registries—npm, PyPI, and Crates.io—comprising over 34 malicious packages and more than 384 related versions. The attack targets developer communities in cryptocurrency, DeFi, Solana, Sui/Move, and AI. It may lead to the theft of cryptocurrency wallets, SSH keys, cloud credentials, GitHub/AWS tokens, browser data, and other sensitive developer information. Some malicious payloads also attempt persistence via mechanisms including `.cursorrules`, `CLAUDE.md`, Git hooks, cron, systemd, and SSH. SlowMist recommends immediately removing affected packages, isolating compromised systems, rotating exposed credentials, rebuilding CI environments and developer machines from clean images, and conducting comprehensive reviews of GitHub, cloud, SSH, and wallet-related activities.
According to research by security firm Socket Security, a cryptocurrency-stealing supply chain attack dubbed “TrapDoor” spans npm, PyPI, and Crates.io, involving over 34 malicious packages and 384 related versions and artifacts. The attack targets cryptocurrency, DeFi, Solana, Sui, Move, and AI developers. Attack samples can steal sensitive information including SSH keys, wallet data, AWS credentials, GitHub tokens, browser data, and environment variables. Specifically, npm packages execute the shared payload `trap-core.js` via the `postinstall` hook; PyPI packages execute remote JavaScript upon import; and Crates.io packages steal local keystores via `build.rs`. Socket has flagged all related packages as malicious and reported them to the respective package registries.
Wasabi Protocol released a security incident update, stating that the attacker exploited a Spring Boot Actuator configuration vulnerability in its AWS infrastructure to steal private keys controlling EVM smart contracts, and subsequently drained approximately $4.8 million in user funds and $900,000 from the protocol’s treasury—totaling roughly $5.7 million in losses. The attack chain originated from a public-facing analysis server whose Actuator heap dump was not properly password-protected, enabling the attacker to obtain credentials for another server and ultimately gain control of the smart contract private keys. This incident affected only EVM deployments—including certain treasuries on Ethereum, Base, Blast, and Berachain—while Solana deployments and the Prop AMM remained unaffected. No final user compensation plan has been announced yet; however, “ensuring all affected users are compensated” remains the team’s top priority. Updates on the investigation will be shared with the community via Discord.
Vercel CEO Guillermo Rauch (@rauchg) announced that Vercel is conducting an in-depth investigation into the April 2026 security incident. The investigation revealed that the attackers initially breached Vercel’s systems via Context.ai’s account—a startup—but their activities extended far beyond this initial intrusion. Threat intelligence indicates that the attackers distributed malware to steal Vercel account credentials and API keys from other service providers, then used those keys to rapidly and extensively enumerate non-sensitive environment variables. To trace the root cause, Vercel has processed nearly 1 petabyte of network and API logs. Vercel is collaborating with industry partners—including Microsoft, AWS, and Wiz—to respond jointly and has proactively notified other potentially affected parties, urging them to rotate credentials and adopt security best practices.
According to an official incident post-mortem report on the CoW Swap attack, its domain cow.fi was compromised via a supply-chain attack on April 14, 2026. Attackers exploited social engineering tactics to infiltrate the .fi domain registration process and hijack DNS resolution, causing users attempting to access swap.cow.fi to be redirected to a phishing site for several hours. During this period, attackers deployed a counterfeit trading interface and attempted to trick users into connecting their wallets and signing malicious transactions. The report states that this incident did not impact CoW Protocol’s on-chain smart contracts, backend systems, or user fund security; core infrastructure—including services hosted on AWS and Vercel—remained uncompromised. The attack occurred exclusively during the domain registration and transfer process: attackers gained control by forging identity documents and exploiting vulnerabilities in the registration workflow, briefly modifying the domain’s DNS records. The team detected the anomaly within 19 minutes and initiated emergency response procedures, subsequently migrating to cow.finance and fully restoring the cow.fi domain within approximately 26 hours. CoW’s team noted that affected users were primarily those who visited the official website during the domain hijacking window. Preliminary estimates place losses at around $1.2 million. The cow.fi domain has since been reactivated with enhanced security measures—including RegistryLock—and the team has launched external security audits, legal proceedings against the perpetrators, and is developing a potential user compensation plan. The official statement emphasizes that the vulnerability has been patched and outlines plans to improve domain infrastructure security through governance initiatives and industry collaboration.
According to Cointelegraph, researchers from the University of California recently revealed security risks in certain third-party AI large language model (LLM) routers that could lead to the theft of cryptocurrency assets. The study found that LLM routers—acting as API intermediaries—can read plaintext information; some routers were discovered injecting malicious code and stealing credentials. The research team tested 28 paid and 400 free routers, identifying nine routers that actively injected malicious code, two that deployed trigger-avoidance mechanisms, and 17 that accessed Amazon Web Services (AWS) credentials. One router even transferred ETH using the researchers’ Ethereum private key. The study notes that malicious behavior by routers is difficult to detect, and the “YOLO mode” present in some AI agent frameworks—which automatically executes commands—further increases security risks. Researchers recommend that developers avoid transmitting private keys or mnemonic phrases through AI agents and urge AI companies to implement cryptographic signing of responses to enhance security.