🛡️ CISO Intel — Saturday, 04-04-2026

Due to API rate limiting, here is today’s briefing:

Alright, team. Saturday briefing. Grab your coffee, it’s not a quiet one. Ransomware gangs are working overtime, and AI is making them faster.

🔴 Critical Threats & Active Exploitation

• Google Chrome — CVE-2026-5281 (CVSS High, details withheld) — Use-after-free in the Dawn WebGPU component. Actively exploited in the wild for RCE via crafted HTML. Patch your browsers, yesterday.
• Apple iOS/iPadOS — No CVE-ID given, but MIIT reports active exploitation. Attackers are using SMS, email, and poisoned webpages to implant remote control Trojans, steal data, and gain full privileges on iOS 13.0 to 17.2.1 devices. Update your damn iPhones.
• Progress ShareFile Storage Zones Controller — CVE-2026-2699 (CVSS 9.8) & CVE-2026-2701 (CVSS 9.1) — Chained authentication bypass and arbitrary file upload leading to remote code execution. No confirmed active exploitation yet, but this is a “perfect 10” exploit chain waiting to happen. Patch immediately.

🛡️ CVEs Worth Your Attention

• CVE-2026-26314, CVE-2026-26315 | Sonic Blockchain Client | CVSS not specified | Patches for Go-Ethereum dependency vulnerabilities. | PoC: No

⚡ New TTPs & Attack Research

• Akira Ransomware is hitting new speed records. Halcyon reports attack chains from initial access to encryption in under four hours, some within an hour. They’re using zero-days and intermittent encryption. This isn’t your grandpa’s ransomware. (T1078, T1190, T1486, T1567.002).
• ⚡ AI-Driven Ransomware Exfiltration is 100x faster than human-led attacks in testing. If you’re not ready for hyper-speed data theft, you’re already behind.
• ⚡ Claude Code Leak Exploitation: Threat actors are weaponizing the accidentally leaked Anthropic Claude Code source tree (from npm) to distribute Vidar infostealer and GhostSocks proxy via fake GitHub repos. Classic supply chain poisoning, but with AI bait. (T1195.002, T1566.001).
• ⚡ Poisoned Axios NPM Package: Malicious versions (1.14.1, 0.30.4) of the widely used Axios HTTP library are delivering cross-platform RATs after attackers stole maintainer credentials. Linked to North Korean group UNC1069. Check your dependencies, now. (T1195.002, T1071.001, T1574.006).
• ⚡ New Android Rootkit: Operation NoVoice: McAfee found this in 50+ Google Play apps (2.3M downloads). It uses older Android vulns (2016-2021) to install a persistent rootkit that survives factory resets, injects code, and steals data. (T1574.011, T1059.006, T1555.003).
• 🧪 Mirax Android Banking Trojan: A new Malware-as-a-Service (MaaS) offering remote control and overlays for 700+ financial apps, selling for $3,000/month. The barrier to entry for mobile banking fraud just dropped lower.
• 🧪 CrystalRAT MaaS: Another new MaaS, aggressively promoted on Telegram and YouTube. It’s a “one-stop shop” for attackers: Chromium credential harvesting, remote audio/video capture. (T1555.003, T1113).
• ⚡ Next.js Credential Harvesting: Large-scale operation exploiting CVE-2025-55182 in Next.js apps. Automated scripts are pulling database creds, SSH keys, and API keys from 766+ hosts. Cluster UAT-10608 is behind this, using a “NEXUS Listener” web GUI. (T1552.001, T1003.008, T1078).
• ⚡ Malicious LNK files & GitHub in S. Korea: Windows users targeted with LNK files triggering multi-stage malware, leveraging GitHub for hosting. Basic, but effective. (T1204.002, T1102.001, T1566.001).
• 🛡️ ClickFix Attacks on macOS: Apple’s added new warnings to Terminal against social engineering attacks tricking users into pasting malicious code. Good, but users are still the weakest link.

🏗️ DevSecOps & Cloud Security

• Supply Chain Attacks on AI Packages: Q1 2026 saw a sharp surge in credential stealers and malware targeting AI developer tools. This isn’t just theory; it’s happening, compromising LiteLLM, Telnyx SDK, and Axios via poisoned CI/CD pipelines (e.g., Trivy).
• Vertex AI Vulnerability: “Double Agents” can weaponize Google Cloud AI for data theft, with breaches in ~72 minutes. Identity weaknesses are the root cause. Focus on BYOSA, environment isolation, and continuous IAM monitoring for Vertex AI service accounts. (T1078.004, T1530).
• AI-Related Data Breaches: GitGuardian’s 2026 report shows an 81% surge in AI service credential leaks. Unmonitored generative AI use is a massive data exfiltration risk.

🔧 Patches & Vendor Releases

• Google Chrome — Version 146.0.7680.177/178 (Windows/macOS), 146.0.7680.177 (Linux) — Addresses CVE-2026-5281 and 20 other vulnerabilities. 🟢 solid fix. Apply it.
• Progress ShareFile — Security updates available for Storage Zones Controller. Addresses CVE-2026-2699 and CVE-2026-2701. 🟢 solid fix. Prioritize this.
• Apple iOS/iPadOS — Updates for iOS 13.0 to 17.2.1. 🟢 solid fix. Don’t delay.
• Apple macOS — New Terminal warnings against ClickFix. 🟡 partial/workaround needed. It’s a warning, not a block. Training still critical.
• Sonic Blockchain Client — Updates to Go-Ethereum (Geth) version 1.16.9, patching CVE-2026-26314 and CVE-2026-26315. 🟢 solid fix.

🧪 Threat Intel & Malware

• ShinyHunters Claims Cisco Breach: The group claims to have breached Cisco via Salesforce CRM, Salesforce Experience Cloud, and AWS S3, alleging 3M+ records and internal GitHub repos stolen. Cisco hasn’t fully confirmed, but this is a serious claim.
• Akira Ransomware is a top threat, noted by FBI/CISA. Their speed and double-extortion model make them highly effective.
• Play Ransomware targets Brokk: Russia-linked Play ransomware claims 4GB data exfiltration from Swedish manufacturer Brokk (financials, payroll, client files). They’re threatening to leak if not paid. They’ve hit over 1,100 orgs.
• SparkCat Malware Variant: Found on both Apple App Store and Google Play, this mobile malware is stealing crypto wallet recovery phrases from photo galleries. iOS version targets English, Android targets Japanese, Korean, Chinese keywords. Your mobile devices are not safe.
• UNC1069: The poisoned Axios NPM package is linked to this North Korean threat group. Nation-states are actively targeting open-source supply chains.

🌐 Industry, Brand & Internet Security

• Shwapno Data Breach: Bangladeshi company “Shwapno” suffered a data breach, highlighted as a “wake-up call for businesses.” Another day, another breach.
• Nacogdoches Memorial Hospital Hack: Over 257,073 patients’ personal and medical info stolen in January. Healthcare continues to be a soft target.
• Italy Fines Intesa Sanpaolo $36M: Italian DPA fined the country’s largest bank for failing on technical safeguards, leading to an employee illegally accessing thousands of accounts for over two years. Regulatory hammer coming down on lax internal controls.
• FBI Wiretap Infrastructure Breach: Classified as a major incident, state-sponsored Chinese hackers suspected. Compromised system held “returns from legal process” and PII of FBI investigation subjects. Geopolitical cyber friction continues.
• Meta Pauses Work with Mercor: A data breach at AI data vendor Mercor potentially exposed AI industry secrets and model training data, causing Meta to halt work. Supply chain risk extends to your AI vendors.
• “Theranos Playbook” in Cybersecurity: An article highlights how hype and bold claims can overshadow measurable value in our industry. Don’t fall for the BS. Demand proof of outcomes.
• Japan, Singapore, S. Korea AI & Data Security: These nations are enhancing domestic controls over AI and data security to reduce reliance on foreign tech. The push for digital sovereignty is real and will impact global supply chains.

📋 Compliance & Regulatory

• AILeakMonitor.com Launch: New platform specifically for AI data leak prevention, designed for regulated industries (HIPAA, banking, PII). This is a direct response to growing regulatory pressure around generative AI.

💡 Marcus’s Take

The signal is clear: AI is no longer just a target; it’s an accelerant. Attackers are weaponizing AI and exploiting supply chains at unprecedented speeds, making yesterday’s “fast response” today’s “too late.” If your security strategy isn’t obsessively focused on identity, supply chain integrity, and real-time behavioral detection across every layer—especially cloud and AI—you’re going to be left with a blast radius you can’t contain. Trust nothing, verify everything, and pressure your vendors to deliver outcomes, not just promises.