Threat Intel

• AI Hacking: Attack Paths, Real CVEs and Defensive Controls

  AI hacking covers AI-assisted intrusions and attacks on models, agents and ML supply chains, mapping attack paths, a PyTorch RCE and defensive controls.

• Bulletproof Hosting: Why Takedowns Keep Failing

  Sanctions and server seizures hit bulletproof hosting through 2025 and 2026, yet the infrastructure keeps returning. What actually degrades it.

• Standard Chartered's CISO on AI's Dual Role in Bank Security

  Standard Chartered group CISO Cezary Piekarski says AI now accelerates bank defenses and adversary tradecraft alike, and CISOs must speak business terms.

• Hotel Wi-Fi Hijack Uses Fake Updates to Deliver Spyware

  Microsoft attributes CaptiveCrunch to Storm-2945, a Midnight Blizzard sub-cluster using hotel captive portals to deploy malware and steal cloud tokens.

• H96 Android TV Boxes Ship With Ad-Fraud and Proxy Malware

  Bitsight traced 38,000 H96 Android TV boxes shipping with pre-installed malware that rents the owner's IP as a proxy and runs ad fraud when idle.

• UAC-0145 Uses ClickFix CAPTCHAs to Deploy Malware in Ukraine

  UAC-0145 ran ClickFix CAPTCHA lures on 10 or more compromised sites in mid-2026, tricking Ukrainian targets into running PowerShell that stages malware.

• May 2026 Security Roundup: Cisco, Exchange, and Exim

  The May 2026 security roundup covers exploited Cisco SD-WAN and Exchange flaws, a critical Exim bug, NGINX and Defender issues, and Patch Tuesday.

• Remediation Programs: Why Fix Validation Matters

  Mandiant M-Trends 2026 puts mean time to exploit at negative seven days while Verizon's 2025 DBIR finds edge devices take 32 days to remediate.

• AI Security Threat Intelligence Hub

  Tech Sentinel's AI security hub: agentic AI threat models, a PyTorch model-loading RCE, AI fraud detection limits, and supply-chain fraud cases.

• AI Agents Rewrite the Threat Model for Security Teams

  Clinejection, the AI-augmented FortiGate campaign, and the OpenClaw exposure wave show why agentic AI is best treated as an insider-threat problem.

All tags · Browse all topics