SlowMist disclosed KREMLIN, a Brazilian banking malware operation active since May 2025, which uses multi-stage loaders and malicious browser extensions to steal credentials and data. The malware bypasses Chromium security mechanisms and leverages Ethereum smart contracts as dead-drop resolvers for C2 infrastructure, with 1,515 infected hosts primarily in Brazil.
At GISEC, a major AI and cybersecurity conference, attendees highlighted industry concerns about malicious code in AI assistants and supply chain vulnerabilities. Google Cloud Security and Microsoft presented security approaches involving agent approval systems and scanning, which align with eyebrowcc's artifact inventory and control mechanisms designed to prevent unauthorized agent actions in Web2 and Web3 environments.
A developer attending GISEC conference discusses eyebrow, a security tool for monitoring AI agents. The tool inventories agent artifacts, validates content hashes, maps host access, and flags unauthorized changes to prevent malicious code execution in both web2 and web3 environments. Presentations from Google Cloud Security and Microsoft highlighted similar concerns about autonomous exploitation and supply chain risks.
A post from PF Mental Health Hotline discusses 'The Minute Market and the Mind,' a topic linking market dynamics to mental health considerations.
A user warns about multiple fraudulent accounts impersonating OnRe Finance on X, directing victims to fake wallet-connect sites designed to drain cryptocurrency assets. OnRe Finance's official account confirms these impersonations and advises users to only trust verified channels.
Researchers at UC Santa Barbara discovered that LLM API routers used for cost optimization can act as malicious intermediaries, with 9 routers actively injecting malware into AI responses and 17 stealing credentials. The risk is amplified when autonomous agents execute code without human approval, potentially allowing attackers to compromise systems through compromised routers or prompt injections.
A security researcher highlights vulnerabilities in Uniswap's approval process, demonstrating how a malicious hook extracted 4.8% of a transaction's value from a swap routed through the platform's API, illustrating risks when aggregators cannot adequately filter or validate hooks.
Anthropic's threat report documents hostile actors from Yemen, China, Russia, and Iran using Claude AI for weapons development, including guidance systems for missiles and drones. The most serious case involved a Yemeni cell using Claude Code to develop software for guided rockets and ballistic missiles, employing evasion tactics to circumvent safeguards, though no operational weapons were successfully fielded.
A user discovered concerning fine print after being asked to provide their work email, raising security or privacy concerns about data handling practices.
A post about persistent AI risks was shared on X, garnering engagement from The AI Investor account.
Directors Yuval Abraham and Rachel Szor received the Venice Film Festival's Special Jury Prize for NAZA, a documentary alleging the Israeli military uses AI to target civilians in Gaza based on anonymous testimony from 24 claimed former intelligence officers. The film's claims about a 500-casualty strike and autonomous targeting systems are disputed by the IDF and fact-checkers, with casualty records and military documents contradicting the film's core allegations.
A ZKX Wallet security post warns about phishing approvals where malicious dApps or compromised front-ends trick users into signing transactions that grant unlimited token spend permissions or transfer NFTs, disguised as routine wallet prompts. Users are advised to carefully read transaction details rather than blindly clicking through signature requests.
A social media post discusses how major political parties compete for the support of a committed voter base (25-30% of the electorate) willing to vote for candidates perceived as extreme or objectionable, characterizing this dynamic as both illogical and strategically rational from the parties' perspective.
Discussion on X about AI agent engineering in the Sui ecosystem, focusing on signing, permissions, and DeFi execution capabilities.
An RBL Web3 Academy article discusses understanding DeFi before use, posted on X with moderate engagement.
A post explains hardware wallets as physical devices that isolate private keys from internet-connected devices, using examples like Ledger and Trezor. Key points include that crypto remains on the blockchain, hardware wallets add security through required transaction approval on the device, recovery phrases must stay secure, and users must still verify transactions carefully to avoid approving malicious ones.
A social media post discusses an unexplained appearance of $326 million, shared on X with significant engagement.
An attacker exploited a flaw in EtherFi's AtomicQueue contract to steal 15.45 ETH by abusing existing ERC-20 approvals. The vulnerability stemmed from missing access-control checks in the solve() function, allowing an unauthorized address to manipulate the contract into treating the attacker as a legitimate solver and execute unauthorized token transfers.
Two X posts discuss cryptocurrency security threats. User arkilus78 provides wallet protection techniques including transaction simulation, hardware wallet verification, and regular approval revocation. User Oveck shares a personal phishing scam experience where they lost 6.42 ETH through a malicious CollabLand bot on Telegram, emphasizing vigilance even for experienced users.
A social media discussion about a fictional character named Viola whose arc deliberately avoided redemption, keeping her as a morally broken person seeking approval and validation despite her wishes for a different outcome.