Cybersecurity Brief – 2026-05-18

Mon, 18 May 2026 00:00:00 +0000

Major Incidents or Breaches

Grafana has confirmed a data breach following claims by the Coinbase Cartel group, which is reportedly linked to ShinyHunters, Scattered Spider, and Lapsus$ threat actors. The extent of data accessed or stolen has not yet been fully disclosed ^[11].
Historical analysis has revealed that the fast16 malware, a Lua-based tool predating Stuxnet, was specifically designed to sabotage nuclear weapons testing simulations ^[1].

A critical vulnerability in NGINX (CVE-2026-42945) affecting both NGINX Plus and NGINX Open has been publicly disclosed and is now being actively exploited in the wild. The flaw can cause worker process crashes and, if Address Space Layout Randomization (ASLR) is disabled, may enable remote code execution ^[3]^[12].
The “MiniPlasma” Windows zero-day vulnerability allows local privilege escalation to SYSTEM level on fully patched Windows systems. A proof-of-concept exploit has been released publicly ^[2]^[7].
A local privilege escalation vulnerability in the Linux kernel’s rxgk module, known as DirtyDecrypt, now has a public proof-of-concept exploit. Successful exploitation grants root access on affected Linux systems ^[5].

The Coinbase Cartel cybercrime group, with reported links to ShinyHunters, Scattered Spider, and Lapsus$, is implicated in the recent Grafana breach ^[11].
The Tycoon2FA phishing kit has introduced device-code phishing techniques and leverages Trustifi click-tracking URLs to compromise Microsoft 365 accounts ^[8].

AI agents are increasingly capable of autonomously discovering and exploiting obscure vulnerabilities, coinciding with a surge in potentially insecure AI-generated code. This trend is compelling defenders to adapt their security strategies ^[9].
At the Pwn2Own Berlin 2026 contest, researchers demonstrated exploits for 47 zero-day vulnerabilities across Windows, Linux, VMware, Nvidia, and AI products, earning a combined $1.3 million in rewards ^[6]^[13].

South Korea will use its upcoming local elections as a test for new regulations intended to curb the spread of deepfakes, providing an early case study on the efficacy of legal controls against AI-driven disinformation ^[10].