Daily Security

Claude Fable is out. As you can see, the new model claims to have been significantly improved. Looking forward to giving it a shot https://www.anthropic.com/news/claude-fable-5-mythos-5

PASSWORD MANAGER 'BITWARDEN' CLI WAS COMPROMISED IN A SUPPLY CHAIN ATTACK ⚠️ BITWARDEN CLI 2026.4.0 WAS COMPROMISED AS PART OF THE ONGOING CHECKMARX SUPPLY CHAIN CAMPAIGN AFTER ATTACKERS ABUSED A GITHUB ACTION IN BITWARDEN’S CI/CD PIPELINE: SOCKET - INFINITYHEDGE SOCKET: IF YOU USE BITWARDEN CLI, WE RECOMMEND REVIEWING YOUR CI LOGS AND ROTATING ANY SECRETS THAT MAY HAVE BEEN EXPOSED TO THE COMPROMISED WORKFLOW SECURITY RESEARCHER ADNAN: I BELIEVE THIS IS THE FIRST TIME A PACKAGE USING NPM TRUSTED PUBLISHING HAS BEEN COMPROMISED

Drone Security Cheat Sheet — and the attack surface is bigger than you’d expect. Drones aren’t just flying cameras anymore. They’re networked IoT devices with GPS modules, companion computers, Wi-Fi links, and open ports. That means: GPS spoofing, deauth attacks, firmware tampering, and sensor manipulation are all real threats. The key vulnerable endpoints: • Communication links — unencrypted MAVLink or Wi-Fi traffic can be intercepted or replayed • Companion computers — open SSH/FTP ports are low-hanging fruit for attackers • Sensors — GPS, cameras, and altimeters can all be fed false data (think Stuxnet, but airborne) • Physical access — if a drone is captured, unencrypted storage = game over The mitigations read like an IoT security playbook: secure boot chains, firmware signing, encrypted storage (LUKS, gocryptfs), MAVLink 2.0 message signing, WPA3, and 802.11w to block deauth attacks. One interesting note: researchers are exploring watermarked sensor signals — using entropy analysis to detect if sensor values have been tampered with. Full cheat sheet: cheatsheetseries.owasp.org/cheatsheets/Drone_Security_Cheat_Sheet

Interesting story behind Drift Protocol Hack -> On April 1, 2026, the Drift protocol (a DeFi platform on Solana) was exploited in a sophisticated attack that had been planned over six months. -> Starting in Fall 2025, attackers posed as a legitimate quantitative trading firm, meeting Drift contributors in person at multiple crypto conferences across several countries. They built trust through months of genuine-looking engagement — discussing trading strategies, onboarding a vault, depositing over $1M of their own capital, and holding working sessions with team members. -> The actual compromise likely happened through malicious software: one contributor cloned a code repository shared by the group, and another downloaded a TestFlight app they presented as a wallet product. A known VSCode/Cursor vulnerability from late 2025 may have been exploited, which allowed silent code execution just by opening a file. -> Once the exploit occurred on April 1, the attackers scrubbed all their Telegram chats and malicious tools. Drift has since frozen protocol functions, removed compromised wallets from the multisig, and flagged attacker wallets across exchanges. -> The attack is attributed with medium-high confidence to the same North Korean state-affiliated group (tracked as UNC4736/AppleJeus/Citrine Sleet) behind the October 2024 Radiant Capital hack, based on onchain fund flows and operational overlaps. Notably, the people who appeared in person were not North Korean nationals — DPRK groups are known to use third-party intermediaries for face-to-face interactions. Mandiant has been engaged but has not yet formally attributed the attack. The investigation is ongoing. Source: https://x.com/DriftProtocol/status/2040611161121370409

Meet the Cybersecurity Baron: a specialized LLM fine-tuned for offensive security. This isn't your average chatbot. It's a quantized, 6-bit GGUF model built on Llama 3.1 Instruct, designed to think like a penetration tester. https://huggingface.co/AlicanKiraz0/Cybersecurity-BaronLLM_Offensive_Security_LLM_Q6_K_GGUF

NVIDIA Unveils NemoClaw: The Missing Security Layer for AI Agents OpenClaw exploded out of nowhere in January. An Austrian developer named Peter Steinberger built the first version in about an hour, and within weeks it became one of the fastest-growing open source projects in GitHub history- outpacing Linux's 30-year download record in just three weeks. The PROBLEM with autonomous agents OpenClaw's strength - broad, unchaperoned access to your system - is also its fundamental risk. Early versions had well - documented vulnerabilities around prompt injection and unconstrained file access. Most got patched, but no software fix can resolve the structural tension between an agent that needs wide access to be useful and an organization that can't afford to let AI roam freely through production systems. NVIDIA's answer, announced at GTC this week, is NemoClaw. What NemoClaw actually does? NVIDIA NemoClaw is an open source stack that adds privacy and security controls to OpenClaw. With one command, anyone can run always-on, self-evolving agents anywhere. NemoClaw uses NVIDIA Agent Toolkit software to secure OpenClaw. It installs NVIDIA OpenShell to enforce policy-based privacy and security guardrails, giving users control over how agents behave and handle data. It also evaluates available compute resources to run high-performance open models like NVIDIA Nemotron™ locally for enhanced privacy and cost efficiency. What's the catch? It's early. NVIDIA is describing NemoClaw as "alpha" and explicitly warns developers to expect rough edges. The sandbox orchestration isn't production-ready yet. The starting point, as they put it, is just "getting your own environment up and running." Also worth noting: Peter Steinberger, OpenClaw's creator, was recently acquihired by OpenAI. The project is now managed by a foundation to ensure vendor - neutral governance - but how that plays out as NVIDIA builds commercial tooling on top remains to be seen. Why this matters The McKinsey stat floating around is that 80% of organizations deploying AI agents have encountered "risky or unexpected behavior." The Alibaba incident from earlier this month - where an AI agent spontaneously started mining crypto and punching holes through firewalls - underscored just how real those risks are. NemoClaw doesn't solve the fundamental challenge of aligning autonomous systems with human intent. But it does provide the infrastructure layer that lets enterprises set boundaries, enforce policies, and actually audit what their agents are doing. Whether that becomes the industry standard or just one option among many will depend on adoption. But NVIDIA is betting big that when it comes to agent trust, enterprises will want to buy infrastructure rather than build it themselves. Useful Materials: • Good post from Kirill • You can try Nvidia Solution Here • Initial Nvidia Presentation Here

The AI Skills Explorer is a searchable database of AI skill files- structured prompts in YAML or markdown format-designed specifically for smart contract security auditing. These skills come from top security teams and researchers in the space, giving you battle-tested configurations you can drop directly into your workflow. https://smartcontractshacking.com/tools/ai-skills-explorer Follow for more @ethers_security

I would call it a good beginning if you're considering to build a sec tool with AI superpowers Why it is worth a read 🌟 The article presents Amy, a semi-autonomously evolving, freely available open-source specialist smart contract auditor focused on Vault/ERC4626 protocols. While Amy can work with any AI platform, she self-evolved using Claude, which showed the best performance in their testing. What Amy can actually find 🥇 In internal testing on Cyfrin private audits, Amy has found Critical, High, Medium, and Low severity issues — some of which a Junior Auditor would be unlikely to catch. She can also find more complex bugs by defining a list of protocol invariants and attempting to break them, and can make recommendations to strengthen codebase defensiveness, something Junior Auditors typically don't do The bigger vision By open-sourcing Amy's Primer document, any auditor or developer can use and evolve her themselves, paying only the Claude compute cost. The author's hope is that the smart contract security community will embrace AI and collaboratively evolve an open-source collection of freely available specialist AI auditors. Source: https://dacian.me/using-claude-to-evolve-specialist-ai-smart-contract-auditors Amy: https://github.com/devdacian/ai-auditor-primers How to use Amy effectively? https://x.com/DevDacian/status/1945790783493423525?s=20

Good start https://dacian.me/using-claude-to-evolve-specialist-ai-smart-contract-auditors

OpenSourceMalware.com is a community-driven threat intel platform for tracking malicious packages across npm, PyPI, GitHub & more. Browse verified threats, share findings, and protect your dependencies🔒🤝 👉 https://opensourcemalware.com

First high-severity Solidity compiler bug since 2016 — identified by Hexens Hexens has identified a HIGH severity vulnerability in the Solidity compiler itself. Not in a specific project. In Solidity. The issue, TSTORE Poison, can silently corrupt contract storage and introduce critical vulnerabilities without obvious indicators. This makes it particularly dangerous for developers and auditors who rely on compiler-level guarantees. To assess the ecosystem-wide impact, we used Glider to scan integrated chains and evaluate the potential blast radius. Cross-chain impact analysis at this scale is exactly what Glider is designed for. 🔎 Technical write-up: https://hexens.io/research/solidity-compiler-bug-tstore-poison 📢 Official announcement: https://x.com/solidity_lang/status/2024181697168945228?s=46 If you develop or audit Solidity smart contracts, we strongly recommend reviewing the report

Aave V4 audit just got released by Trail Of Bits Available here

Exploring the Security Risks of AI Assistants like OpenClaw https://www.hiddenlayer.com/research/exploring-the-security-risks-of-ai-assistants-like-openclaw

🛡️ New protection against overflow errors in Go By default, Go ignores overflow in integer arithmetic operations, which hides serious vulnerabilities. The go-panikint project is a modified Go compiler that turns such overflows into explicit panics at runtime. This helps to find real bugs, like in Cosmos SDK RPC pagination, revealing previously invisible issues during fuzzing. 🔗 Details

🚨 JUST IN: Sui reports a mainnet outage, with the network currently at a standstill as the core team investigates. News | Markets | YouTube

👏 https://frameworks.securityalliance.org/intro/introduction

Trust Wallet Browser Extension v2.68 Incident https://trustwallet.com/ru/blog/announcements/trust-wallet-browser-extension-v268-incident-community-update