Author: Staff Writer

Fast Facts Singapore’s rapid AI adoption among SMEs exceeds governance capacity, highlighting a need for accountable frameworks like the Trustmark certification to ensure quality and security. Partnerships with Grab and RSM expand SME security and AI literacy, targeting large-scale, practical capability development in security-conscious AI adoption. The launch of Sonar’s AI remediation agent addresses security flaws in AI-generated code, emphasizing the importance of secure-by-design practices in AI development. Singapore’s integrated, coordinated approach to AI adoption, security, and infrastructure—particularly through pilot programs like quantum-safe migration—serves as a global reference model for comprehensive AI governance. Singapore’s Comprehensive Approach to AI Governance Singapore…

Microsoft introduces advanced security tools like MDASH and integrated workflows to detect, validate, and remediate vulnerabilities early in the development process, enhancing proactive security. New capabilities enable secure, compliant development of AI agents from inception, with built-in governance, runtime controls, and comprehensive visibility across environments. Microsoft enhances data and model security, including runtime data loss prevention and AI model scanning, to ensure trust and prevent vulnerabilities before deployment. The holistic approach embeds security throughout the AI development lifecycle, emphasizing trust, safety, and streamlined governance to support rapid innovation responsibly. Integrating Security Into Daily Development and Operations In today’s fast-paced tech…

Summary Points Ransomware operators are making costly mistakes, such as hardcoding recovery keys or discarding private keys, which hinder victim recovery efforts and reduce their leverage. Despite the general illegality of cybercrime in CIS countries, local authorities often provide safe harbor, allowing cybercriminal groups to operate with minimal interference, especially if they avoid in-country targets. Certain ransomware groups explicitly prohibit targeting CIS organizations, highlighting strategic operational boundaries and potential consequences of attacking in-region entities. Threat, Attack Techniques, and Targets The threat involves a ransomware affiliate program called Nova, linked to the larger RAlord crew. This group conducts ransomware attacks mainly…

Quick Takeaways Kali365 has expanded from a Microsoft-focused phishing tool to a broader platform targeting AWS, Okta, Russian online services, and others, including MAX Messenger with over 80 million users. It employs device code phishing that hijacks device authorization workflows, enabling attackers to access accounts without stealing credentials, bypassing MFA defenses. Arctic Wolf’s analysis revealed Kali365’s evolving infrastructure, impersonating numerous platforms and posing a significant threat to enterprises globally. Experts recommend comprehensive security awareness training and specific detection measures, as device code phishing kits like Kali365 are rapidly proliferating across multiple environments. Kali365 Expands Its Reach to Broader Targets Recently,…

Quick Takeaways The Russian group Gamaredon exploits CVE-2025-8088 in WinRAR using HTML applications and VBScript downloaders to deliver malware, including the GammaWorm persistent backdoor and GammaSteel data-stealer, targeting Ukrainian and global entities. GammaWorm manipulates NTFS Alternate Data Streams and legitimate platforms like Telegram for covert C2 communications, enabling long-term espionage and data exfiltration. Advanced modular malware like GammaLoad and its variants demonstrate highly obfuscated, adaptable attack chains that can deploy multiple payloads, including destructive wipe tools, with potential for future reuse. Threat, Techniques, and Targets The threat is from a Russian hacking group called Gamaredon. They are actively exploiting a…

Quick Takeaways Mustang Panda conducted a sophisticated cyberattack utilizing a multi-layered chain, disguising malicious files as legitimate updates to evade detection. The malware chain involved legitimate signed binaries, encrypted payloads, and memory-based execution techniques, complicating static and dynamic analysis. Once installed, the PlugX RAT communicated over HTTPS to command servers, carrying extensive capabilities such as file management, process control, and stealth operations. Security experts recommend monitoring specific file names, registry entries, and behavioral indicators across different directories for effective detection of this stealthy attack. What’s the Problem? A notorious Chinese state-sponsored hacking group, Mustang Panda, launched a highly sophisticated cyberattack…

Fast Facts Anthropic has broadened access to its AI-driven cybersecurity tool, Claude Mythos Preview, to around 150 new organizations across over 15 countries, primarily critical infrastructure providers. Early partners identified over 10,000 high-severity security flaws, demonstrating Mythos-class AI’s powerful vulnerability detection capabilities in enterprise codebases. The expansion enhances layered security functions, including patching, threat detection, penetration testing, and legacy code rebuilding, amid industry concerns about AI misuse. Anthropic aims for wider access in the future, emphasizing safeguards and evolving towards a comprehensive cybersecurity model that shifts AI from vulnerability detection to full-cycle defense and remediation. The Issue Anthropic has greatly…

Essential Insights CISA warns of active exploitation of CVE-2024-21182, a critical vulnerability in Oracle WebLogic Server, added to its KEV list on June 1, 2026, highlighting urgent risks to enterprise systems. The flaw allows remote, unauthenticated attackers to gain unauthorized access via network protocols like T3 and IIOP, risking data breaches and potential full system compromise. Exploitation is widespread due to WebLogic’s exposure, with threat actors likely to leverage it for ransomware and financially motivated attacks, especially on misconfigured or internet-facing servers. Organizations must promptly apply Oracle patches, restrict network access to WebLogic services, and enhance monitoring to prevent compromise,…

Fast Facts AI-powered spoofing has increased biometric fraud attempts by 230%, enabling attackers to bypass fingerprint, facial, and voice recognition systems. Generative AI-driven phishing and deepfake attacks surged over 1,300%, causing average breach costs exceeding $5.7 million per incident. Shadow AI practices in organizations drive breaches costing up to $4.63 million, highlighting the risks of unsanctioned AI use and inadequate governance. Threats, Attack Techniques, and Targets AI has significantly changed cyber threats for SMBs. Cybercriminals now use AI to launch more advanced attacks. One major threat involves AI-powered spoofing of biometric data. Attackers can reconstruct fingerprint patterns from high-resolution photos,…

Fast Facts The U.S. CISA has classified CVE-2026-0257, a critical Palo Alto Networks PAN-OS authentication bypass flaw, as actively exploited in real-world attacks, enabling unauthorized VPN access and internal network compromise. The vulnerability allows remote attackers to bypass security measures without credentials, risking data exfiltration, lateral movement, and further system infiltration within enterprise networks. CISA urges immediate patching or application of mitigation steps, and recommends monitoring VPN activity and investigating anomalies to detect exploitation attempts early. The exploit underscores the critical need for proactive patch management and network monitoring as edge infrastructure vulnerabilities increasingly attract threat actors. What’s the Problem?…