Monday 07 September 2026 | Join Free | Upgrade

Hi there, this is your daily ☕️ Cyberpresso.

In today's Cyberpresso:

🔓 Hackers drain $320M from Bitcoin sidechain

💥 Berlin cyberattack leaks sensitive data across the dark web

🛡️ Router flaw grants full control

🛒 Magento zero-day used to backdoor stores

🛠️ N-able patches perfect-10 remote code flaw

Plus: 💡 6 strategies & tactics, 🎁 8 other news you might like, 🧰 6 tools, and 📚 5 papers.

10x the context. Half the time.

Speak your prompts into ChatGPT or Claude and get detailed, paste-ready input that actually gives you useful output. Wispr Flow captures what you'd cut when typing. Free on Mac, Windows, and iPhone.

🔓 Hackers drain $320M from Bitcoin sidechain LINK
  • Purported white hat hackers have drained just under 4,000 Bitcoin worth $319 million from Blockstream's Liquid Network sidechain, prompting the team to pause bridge nodes and tell exchanges to halt LBTC deposits and withdrawals while it investigates.
  • The funds were withdrawn via the SideSwap Peg-out Authorization Key (PAK), though Liquid says that key was not compromised, nor were any others, leaving open whether 11 of 15 functionaries signed off or the withdrawal whitelist failed to hold.
  • The federation wallet's balance fell from 4,200 BTC to 207.275 BTC, and analyst DBCrypto noted the coins are sitting unmixed on Bitcoin, more consistent with a whitehat extraction, while an OP_RETURN message read: "we are whitehats. contact us on chain."
💥 Berlin cyberattack leaks sensitive data across the dark web LINK
  • The ransomware group Rhysida has published nearly six terabytes of data, 1,439,893 files, stolen from Berlin's state administration on the dark web after the city refused to pay the ransom.
  • The hackers had demanded 30 Bitcoin, around two million euros, and set a countdown that expired at about 3.35 pm on Friday; the Berlin Senate said it does not give in to blackmail, and the data was released shortly after.
  • Leaked files include a folder on chemical, biological, radiological and nuclear threat planning, LKA investigation documents, national defense plans, plus personal data on state civil servants like birth certificates, phone numbers and home addresses.
🛡️ Router flaw grants full control LINK
  • Attackers are actively exploiting a flaw in MikroTik RouterOS that hands any remote, unauthenticated attacker a direct shell on the device, opening the door to full network takeover, with MikroTik confirming the bug on September 3, 2026.
  • The flaw sits in a core library shared by multiple RouterOS services and is tied to SSH, so any router with SSH reachable from the internet can be taken over without stealing credentials or any user interaction, regardless of password or key-based login.
  • Fixes shipped across every channel including 7.24.2, 7.23.4, and 6.49.21; upgraded devices now auto-inspect their config at startup and set a "Flagged" status on signs of tampering, but MikroTik urges auditing every device, rotating passwords, and restricting SSH regardless.
🛒 Magento zero-day used to backdoor stores LINK
  • A zero-day flaw called StyleSmuggler in Magento Open Source and Adobe Commerce is being used to fully take over online stores, with e-commerce security firm Sansec disclosing it early on September 5, 2026 because sites are being compromised right now.
  • The flaw is actively exploited by unauthenticated attackers to run code with no login needed, hitting every current version including 2.4.9 and even a fully patched 2.4.6-p15 store, and Adobe has issued no advisory, CVE, or fix as of September 6.
  • Attackers slip malicious PHP into a file Magento writes by abusing "styles" in a GraphQL request, then trigger it via a rendered "Payment Transaction Failed Reminder" email; Sansec advises temporarily disabling GraphQL, and checking both var/report and var/log/system.log.
🛠️ N-able patches perfect-10 remote code flaw LINK
  • N-able has shipped an emergency hotfix for a maximum-severity (CVSS 10/10) remote code execution flaw, tracked as CVE-2026-86218, in its N-central remote monitoring and management platform used by IT teams and managed service providers.
  • The bug lets an attacker with no privileges run malicious code on unpatched N-central instances exposed online in easy-to-pull-off attacks; N-able reports no confirmed exploitation in production, though Huntress has flagged it as a potential zero-day.
  • N-able fixed the flaw on Saturday with N-central 2026.3 Hotfix 4 and urges on-premises customers to upgrade immediately, since systems running HF3 stay vulnerable, while Shadowserver tracks nearly 1,500 exposed servers, mostly in the US and Europe.

AI can build faster. Can your team decide better?

AI can draft the PRD and prototype the idea. Jira Product Discovery helps teams decide whether it belongs on the roadmap. Bring feedback and ideas together, prioritize as a team, and keep your roadmap connected to delivery in Jira.

💡 Strategies & Tactics

> 39 New Methods That Compromise Passkey Authentication: Attackers now bypass passkeys by targeting the surrounding software, sync services, and enrollment steps rather than the cryptography, so bind high-value logins to dedicated biometric hardware.
> MCP's new spec turns a planted prompt into a stolen credential: MCP's new stateless design moves security from the protocol to your endpoints, so a planted prompt injection can now steal a valid access credential.
> The Suit Is Flawless, but the Glass Fogs Up: Why Antidetect Browsers Get Caught by the Network, Not the Fingerprint: Antidetect browsers get caught when video-call technology leaks the user's real IP address around the proxy, so test every profile for network leaks before use.
> Flock's AI search tool for police officers has been reverse engineered — here's what it shows: Reverse-engineering Flock's police software exposed AI search tools that can track people by description, revealing surveillance capabilities with no way to verify accuracy.
> ShreadOS destroys your old PC's data more thoroughly than a hammer ever will: ShredOS is a bootable USB tool that overwrites a drive's data so it cannot be recovered, letting businesses safely retire or resell old hardware.
> Official support for Microsoft Office 2021 ends next month but 0patch is here to help: 0patch will sell security patches for Office 2021 for three years after Microsoft's October cutoff, letting cost-conscious users keep old software safely.

Other news & articles you might like

  • Telerik UI Flaws Let Attackers Chain AES-CBC Padding Oracle to Unauthenticated RCE LINK
  • New Linux toolkit found in trojanized HAProxy targeting South Korean organizations LINK
  • Malicious Chrome Extension Can Steal Login Sessions and Turn PCs Into Remote Backdoors LINK
  • Critical ASUS Control Center CVE-2026-75754 Flaw Allows Unauthenticated Root Access LINK
  • Hackers Turn HiveMQ and Element Messenger Into Control Channels for Windows Backdoors LINK
  • Kimsuky Uses OpenCode AI Agent and GitHub PATs in Operation GitPower Attacks LINK
  • Over 5,400 hacked sites serve ClickFix payloads stored on the blockchain LINK
  • New Panzer Ransomware Hits 16 Victims Across 11 Countries With Data Theft and Encryption LINK

🛠️ Trending tools

Kastra: runtime authorization layer that evaluates AI agent actions before execution, blocking unauthorized tool use, prompt injection, and data exposure with sub-millisecond policy checks. LINK
MonoCloud for Startups: handles authentication and Cedar-based authorization across users, APIs, and AI agents, letting you control, audit, and revoke access, free for a year LINK
Playground: a hands-on sandbox for testing prompt injection attacks against AI agents, letting you explore and understand real LLM security weaknesses LINK
cain-agent: an AI penetration testing engineer for authorized security assessments, with a built-in cloud module covering AWS, Azure, GCP, and major Chinese cloud providers. LINK
Agentic OS: a Rust-based Linux binary that provides each entity with its own SQLite database and isolated sandbox environment. LINK
Prized: builds secure internal tools with AI for ops, support, and finance, featuring pre-connected data, access audit trails, and one-click deploy behind company sign-in LINK

📚 Trending research papers

AI agent attacks get more effective the harder an automated attacker keeps probing, showing that security tests for tool-using agents must account for how much effort a determined attacker invests. LINK
Chatbot safety scoring moves beyond pass-fail testing by grading how models handle harmful requests across four threat levels, revealing that bots with identical failure rates actually behave very differently under pressure. LINK
AI-written phishing emails get more convincing with each layer of personal workplace detail, raising the odds a recipient clicks by ~28% per level, though wrong or vague details backfire and trigger suspicion. LINK
Private search speedups let AI systems find similar items without leaking what you searched for or which records you touched, using large secure GPU memory to sidestep the slow protections older systems relied on. LINK
Software supply chain scanners catch only whether a risky component is present, not whether attackers can actually reach and exploit it, leaving companies blind to how one flaw like Log4j spreads downstream. LINK

🎓 Want to master the AI tools we cover every day?

Our AI Academy has 330+ step-by-step tutorials on ChatGPT, Claude, Perplexity, and every tool that matters. No fluff — just practical workflows you can use at work. Try it free for 7 days.

💬 How did you find today's edition?

We read every reply — just reply to this email and let us know how we can improve!

★★★★★  Nailed it
★★★  Average
  Fail

Not subscribed to ☕️ Cyberpresso yet? Subscribe for free