The threat landscape is accelerating into fully AI-augmented offense. Amazon Threat Intelligence has confirmed a Russian-speaking, financially motivated actor breached 600+ FortiGate devices across 55 countries between January 11 and February 18, 2026 — using commercial generative AI tools across every phase of the attack cycle: tool development, targeting, and execution. No zero-days needed. Exposed management ports plus weak single-factor credentials were enough, and AI gave an otherwise unsophisticated actor the scale to exploit them globally. This is the template now. Low-skill floor, AI ceiling, global reach.
On the vulnerability exploitation front, two CVSS 9.9 critical flaws are actively burning. BeyondTrust CVE-2026-1731 has graduated from initial exploitation to full ransomware campaign infrastructure — web shells, backdoors, C2 implants, lateral movement, and data theft confirmed across financial services, healthcare, legal, and higher education in the US, France, Germany, Australia, and Canada. Simultaneously, Roundcube CVE-2025-49113 was weaponized within 48 hours of public disclosure and is now KEV-listed by CISA alongside a companion XSS flaw. The 48-hour weaponization window is the headline stat — defenders have less than two days after a patch drops before active exploitation begins.
The mobile and supply chain vectors are also live. Intellexa's Predator spyware has been confirmed hooking iOS SpringBoard to suppress the mic and camera activity indicators — streaming feeds to operators while showing users a clean screen. On the supply chain side, the Cline CLI 2.3.0 npm compromise silently pushed OpenClaw onto developer machines via a poisoned postinstall script. And PromptSpy, the first confirmed Android malware using Google's Gemini model at runtime, adapts its persistence behavior per-device. The AI-native threat era is not coming — it's here.