The week in cyber:

📎 Add-in supply chain abuse
🧠 AI in attack workflows
🐛 Active zero-days patched
🔐 Privileged access exploits
☁️ Cloud infra hijacks
🤖 Crypto mining botnets
📡 IRC-based C2 returns
🛠️ PoC exploits weaponized
🏭 Defense sector targeting

🔗 Full Weekly Cybersecurity Recap → https://thehackernews.com/2026/02/weekly-recap-outlook-add-ins-hijack-0.html

🛑 A new academic study mapped password recovery attack paths across Bitwarden, LastPass, and Dashlane—testing zero-knowledge designs against a malicious server model.

Researchers identified 25 attack scenarios impacting vault integrity and recovery flows. No active exploitation reported.

🔗 Research scope, attack methods and vendor fixes → https://thehackernews.com/2026/02/study-uncovers-25-password-recovery.html

🔒 Apple’s latest beta brings end-to-end encryption to RCS — but only for Apple-to-Apple chats so far.

It’s part of iOS 26.4 testing alongside new memory protections designed to block spyware at the kernel level.

🔗 Inside Apple’s new security stack → https://thehackernews.com/2026/02/apple-tests-end-to-end-encrypted-rcs.html

⚡ Microsoft warns some firms are embedding manipulative prompts in “Summarize with AI” buttons.

Researchers identified 50+ hidden instructions that push assistants to remember and recommend specific brands — with effects persisting across future chats, often without user awareness.

🔗 Read → https://thehackernews.com/2026/02/microsoft-finds-summarize-with-ai.html

Most “AI for GRC” still stops at task automation — drafting policies or extracting clauses.

As Yair Kuznitsov, CEO of Anecdotes, explains, agentic GRC replaces the workflow itself. Agents collect evidence, evaluate controls, trigger remediation, and maintain audit trails autonomously.

Decision-making is embedded.

🔗 CCM agent execution model → https://thehackernews.com/expert-insights/2026/02/ai-shouldnt-improve-workflows-it-should.html

⚠️ ☠️ 🤖 A trojanized Oura AI connector is being used to spread SmartLoader malware.

Attackers cloned the MCP server, staged fake GitHub contributors, and planted it in trusted registries. The payload drops StealC to steal credentials, wallets, and cloud access.

🔗 Read → https://thehackernews.com/2026/02/smartloader-attack-uses-trojanized-oura.html

Network Detection & Response is now central to SOC workflows—not an add-on.

Testing Corelight’s Investigator showed analysts tracing exploits, reverse shells, and lateral movement in one dashboard, mapped to MITRE and guided by embedded AI.

NDR serves as a force multiplier for mid-tier teams.

🔗 Inside: hunt workflows → https://thehackernews.com/2026/02/my-day-getting-my-hands-dirty-with-ndr.html

Infosec Compliance Now 2026 | Earn 4 CPEs

Registration for the 6th annual Infosec Compliance Now virtual event is live! Attend and earn up to 4 free CPE credits while learning about AI-powered GRC, cyber resilience, continuous control monitoring using automation, and more!

Register → https://thn.news/cyber-risk-event

⚠️ Cloud attacks move faster than investigations.

Wiz experts show how workloads vanish, identities rotate, and logs expire before response even starts.

Minutes — not days — decide outcomes. See how context-aware forensics rebuilds full attack timelines fast.

🔗 See cloud breaches reconstructed step-by-step → https://thehackernews.com/2026/02/cloud-forensics-webinar-learn-how-ai.html

🛑 A firmware-level Android backdoor called Keenadu is being shipped inside signed tablet builds.

Telemetry shows 13,715 users globally encountered its modules. It injects into every app via core system libraries, enabling remote control, data theft, and ad fraud.

🔗 Read → https://thehackernews.com/2026/02/keenadu-firmware-backdoor-infects.html

🤖 AI assistants with web browsing can be weaponized as stealth command relays.

Check Point showed Copilot and Grok fetching attacker URLs and returning commands through normal responses — blending C2 traffic into enterprise use.

🔗 Read details → https://thehackernews.com/2026/02/researchers-show-copilot-and-grok-can.html

⚠️ CISA added 4 actively exploited flaws to its KEV list, including a Chrome zero-day and critical Zimbra SSRF bug.

Attacks range from browser heap corruption to server command execution and worm delivery.

🔗 Exploited CVEs, affected tech, patch timelines → https://thehackernews.com/2026/02/cisa-flags-four-security-flaws-under.html

⚡ Notepad++ pushed a security update after attackers hijacked its updater to deliver malware to select users.

Version 8.9.2 adds a “double lock” verification system & hardens the auto-updater after a hosting breach enabled poisoned updates.

🔗 Read → https://thehackernews.com/2026/02/notepad-fixes-hijacked-update-mechanism.html

🚨 China-linked UNC6201 exploited a CVSS 10.0 (CVE-2026-22769) Dell RecoverPoint zero-day since 2024 using hard-coded credentials.

Access led to Tomcat web shells, BRICKSTORM installs, and newer GRIMBOLT backdoors built to evade detection.

🔗 Read → https://thehackernews.com/2026/02/dell-recoverpoint-for-vms-zero-day-cve.html

88% of AI pilots never reach production.

Security and IT teams stall not on tools—but execution. Intelligent workflows blend automation, AI decisions, and human review to move phishing response, service requests, and vuln tracking from manual to real time.

🔗 Pre-built workflow breakdowns → https://thehackernews.com/2026/02/3-ways-to-start-your-intelligent.html

⚠️ Four popular VS Code extensions expose developers to file theft and remote code execution.

Researchers say 125M+ installs are affected. Flaws in Live Server, Code Runner, and others enable localhost abuse, malicious configs, and code injection—some still unpatched.

🔗 Read → https://thehackernews.com/2026/02/critical-flaws-found-in-four-vs-code.html

Cybersecurity isn’t preparing for storms anymore. Instability is permanent.

AI threats, regulatory pressure, and geopolitical risk now shape how systems are built and run—not just defended.

🔗 What security looks like when volatility is baseline → https://thehackernews.com/2026/02/cybersecurity-tech-predictions-for-2026.html

🚨 There’s an extreme scarcity of skilled GenAI red teamers—experts estimate fewer than 10,000 globally, while one security industry analyst believes there are fewer than 10 capable of tackling frontier AI risks.

📽️ From TELUS Digital's Fuel iX team: Watch Uncharted: The AI Safety & Security Summit on demand to hear from top security leaders and industry experts as they uncover:

• Hidden vulnerabilities in GenAI models
• How automated red teaming is transforming risk detection
• Actionable strategies to safeguard your AI systems

👉 Watch now: https://thn.news/ai-security-summit-tg

Researchers found a stack overflow in Grandstream GXP1600 phones (CVE-2026-2329, 9.3 CVSS).

A crafted API request can overwrite memory and run code remotely—no login required. Post-exploitation includes credential theft and VoIP call interception via rogue SIP proxy.

🔗 Read → https://thehackernews.com/2026/02/grandstream-gxp1600-voip-phones-exposed.html

🛑 Citizen Lab found Cellebrite tools used on a Kenyan activist’s phone while it was in police custody.

Boniface Mwangi’s Samsung device was accessed in July 2025 and later returned without password protection. Full extraction could include messages, files, and passwords.

The findings follow similar cases in Jordan.

🔗 Read → https://thehackernews.com/2026/02/citizen-lab-finds-cellebrite-tool-used.html