Track East Asia cyber, AI, cloud, and infrastructure risk before it becomes an incident.
Researchers developed TWGuard, an LLM safety guardrail optimized for the Taiwan linguistic context, achieving a +0.289 F1 gain over foundation models and a 94.9% reduction in false positive rate compared to the strongest baseline, demonstrating the value of linguistic context in AI safety. Read more
A new research paper argues that current AI safety case methodologies lack rigor by over-relying on alignment community approaches instead of established safety assurance practices from aerospace, nuclear, and automotive sectors, proposing a stronger framework grounded in decades of safety-critical system validation. Read more
A critical security flaw impacting NGINX Plus and NGINX Open Source, tracked as CVE-2026-42945, has transitioned from public disclosure to active exploitation in the wild. The vulnerability is a heap buffer overflow within the ngx_http_rewrite_module that allows unauthenticated attackers to trigger worker process crashes (denial-of-service) or potentially achieve remote code execution (RCE) via crafted HTTP requests. While RCE is largely mitigated by Address Space Layout Randomization (ASLR), honeypot data confirms threat actors are already weaponizing the flaw to disrupt services. This exploitation coincides with a cluster of attacks targeting openDCIM infrastructure management software, reportedly utilizing AI-assisted vulnerability discovery tools. F5 has released patches, and immediate updates are recommended for all NGINX deployments within the affected version range (0.6.27 to 1.30.0). Read more
Nogosee monitored 13 Taiwan listed-company cyber incident disclosures in April 2026. The cleaned month is incident-only: 13 incident records, 0 governance records, and 0 procurement records. This is a MOPS incident-disclosure snapshot for analyst workflow use, not a complete census of Taiwan cyber activity. Read more
A research study identified over 1.2 million exposed Thai National Identification Numbers via search engines, with most originating from government websites, highlighting systemic data protection failures in public sector digital systems. Read more
A practical tutorial for security teams to integrate Singapore Cyber Security Agency (CSA) alerts into regional East Asia cyber risk workflows, focusing on cloud, government, finance, and supply-chain monitoring with actionable steps for review, triage, and escalation. Read more
OpenAI confirmed two employees' devices were breached in the TanStack supply chain attack, leading to credential exposure and precautionary code-signing certificate rotation, with no impact to customer data or production systems. Read more
A researcher published proof-of-concept exploits for two unpatched Windows vulnerabilities: YellowKey, a BitLocker bypass affecting Windows 11 and Server 2022/2025 via WinRE, and GreenPlasma, an incomplete privilege escalation flaw. The exploits work even in TPM-only BitLocker setups, highlighting risks in automatic decryption workflows. Read more
JPCERT/CC is Japan's Computer Emergency Response Team/Coordination Center, issuing alerts and weekly reports on vulnerabilities affecting software and systems used globally. This guide explains what JPCERT/CC alerts contain, their limitations, and how global security teams can integrate them into routine vulnerability monitoring without overinterpreting their scope or urgency. Read more
Use the official KISA/KrCERT vulnerability feed as a primary source for South Korea cyber risk monitoring. This evergreen workflow provides concrete steps for tracking vulnerability notices, vendor risk, public-sector alerts, and regional exposure without implying recency or requiring hard thresholds. Read more