Cybersecurity Brief – 2026-04-21

Tue, 21 Apr 2026 00:00:00 +0000

Major Incidents or Breaches

KelpDAO suffered a $290 million cryptocurrency theft, attributed to North Korean state-sponsored Lazarus Group hackers. The attackers exploited the DeFi platform to drain assets.
Vercel, the creator of Next.js, confirmed a data breach after a threat actor claimed to have stolen data and offered it for sale for $2 million. The breach reportedly stemmed from compromised OAuth tokens used by an employee’s AI tool.
Seiko USA’s website was defaced, with attackers claiming theft of the Shopify customer database and threatening to leak the data unless a ransom is paid.
Bluesky social network experienced a 24-hour disruption due to a sophisticated DDoS attack, with a pro-Iran hacker group claiming responsibility.
Booking.com was reported as a recent victim of a cyberattack, with further details pending.
A British national, identified as a leader of the Scattered Spider cybercrime collective, pleaded guilty in the US to wire fraud and aggravated identity theft related to cryptocurrency theft.

Newly Discovered Vulnerabilities

Over twenty malicious apps impersonating popular crypto wallets (Metamask, Coinbase, Trust Wallet, OneKey, etc.) were discovered on the Apple App Store, primarily targeting users in China. These apps phished for recovery/seed phrases to steal cryptocurrency.
SGLang (CVE-2026-5760, CVSS 9.8): A critical vulnerability allows remote code execution via malicious GGUF model files.
Anthropic Model Context Protocol (MCP): A critical design weakness was disclosed, enabling remote code execution and posing a supply chain risk for AI deployments.
Serial-to-IP converters from Lantronix and Silex were found to contain 20 new vulnerabilities, exposing OT and healthcare systems to potential attacks.
CISA added eight new exploited vulnerabilities to its Known Exploited Vulnerabilities (KEV) catalog, including three affecting Cisco devices.
WhatsApp was found to leak user metadata, allowing attackers to infer limited information about users without direct contact.
Half of the 6 million internet-facing FTP servers were found to lack encryption, exposing users and enterprises to various attacks.

Notable Threat Actor Activity

Lazarus Group (North Korea) linked to the $290 million KelpDAO crypto theft.
Scattered Spider leader pleaded guilty to multiple cybercrimes, including corporate intrusions and cryptocurrency theft.
Pro-Iran hacker group claimed responsibility for the DDoS attack on Bluesky.
The Gentlemen ransomware group was observed using the SystemBC proxy malware botnet (over 1,570 hosts) for bot-powered attacks, facilitating lateral movement and defense evasion.
Threat actors are abusing QEMU, a machine emulator, in campaigns to evade defenses and distribute ransomware and remote access tools.
Microsoft reported increased abuse of Microsoft Teams by threat actors for helpdesk impersonation and lateral movement using legitimate tools.
Reports indicate use of .wav files as a vector for malware delivery.

Trends, Tools, or Tactics of Interest

Third-party tools and compromised OAuth tokens are increasingly leveraged as initial access vectors in breaches, as seen in the Vercel incident.
There is a rise in threat actors exploiting trusted download paths and legitimate collaboration platforms (e.g., Microsoft Teams) for initial access and lateral movement.
The SystemBC proxy is being used by ransomware operators to automate attacks and maintain persistence within victim networks.
QEMU is being leveraged for defense evasion in recent ransomware and RAT campaigns.
Increase in phishing and malicious app distribution on official platforms, such as the Apple App Store, specifically targeting cryptocurrency assets.
Serial-to-IP converters in OT and healthcare environments remain a weak link due to numerous unpatched vulnerabilities.
AI supply chain risks are highlighted by critical vulnerabilities in AI model protocols and concerns about the misuse of powerful AI tools (e.g., Anthropic’s Mythos tool not publicly released over security fears).
Survey data from LevelBlue indicates ongoing concern among security leaders about the human element in cybersecurity, with workforce education seen as insufficient.

Regulatory or Policy Developments

The US Senate extended surveillance powers under a controversial program until 30 April 2026, following legislative delays.
CISA set federal deadlines in April–May 2026 for remediation of newly catalogued exploited vulnerabilities.

Cybersecurity Brief – 2026-03-18

Wed, 18 Mar 2026 00:00:00 +0000

Major Incidents or Breaches

UK Companies House disclosed a vulnerability that exposed details of millions of firms, which could have been exploited to obtain sensitive company information and alter official records.
Robotic surgery manufacturer Intuitive confirmed a cyberattack after an employee fell victim to a phishing scheme, resulting in unauthorised access to internal business applications.
GlassWorm malware resurfaced in a supply-chain attack targeting over 400 code repositories and packages on GitHub, npm, VSCode, and OpenVSX, impacting the software supply chain.
Sears exposed AI chatbot phone calls and text chats to the public web, allowing access to customer conversations containing contact information and personal details.
LeakNet ransomware adopted the ClickFix technique via compromised websites for initial access and deployed a Deno-based in-memory loader, increasing stealth in attacks.
The RondoDox botnet escalated its activity, targeting 174 vulnerabilities with up to 15,000 exploitation attempts per day.
A phishing attack targeted cybersecurity firm Outpost24, using trusted brands and domains in an attempt to compromise C-suite credentials; the attack was unsuccessful.

Newly Discovered Vulnerabilities

Apple released its first Background Security Improvements update to patch a WebKit vulnerability (CVE-2026-20643) affecting iOS, iPadOS, and macOS, which enabled same-origin policy bypass.
A critical unpatched vulnerability (CVE-2026-32746) in the GNU InetUtils telnet daemon (telnetd) allows unauthenticated remote code execution as root via port 23.
Multiple AI platforms, including Amazon Bedrock, LangSmith, and SGLang, were found vulnerable to data exfiltration and remote code execution attacks using DNS queries.
CISA flagged the year-old Wing FTP vulnerability (CVE-2025-47813) as actively exploited, which can disclose the full local installation path of the application.
A new font-rendering attack was identified that hides malicious commands from AI tools by embedding them in HTML, evading detection by AI-based security solutions.

Notable Threat Actor Activity

North Korean threat actor group Konni was observed deploying EndRAT through phishing campaigns and leveraging the KakaoTalk desktop app to propagate malware to victims’ contacts.
LeakNet ransomware group is now using ClickFix for social engineering and Deno runtime for in-memory malware loading, enhancing stealth and evasion.
The Warlock ransomware group demonstrated augmented post-exploitation activities, including a new Bring Your Own Vulnerable Driver (BYOVD) technique and advanced cross-network movement.
Attackers are increasingly using IPv4-mapped IPv6 addresses to obfuscate attack origins, as noted in recent web scanning activity.

Trends, Tools, or Tactics of Interest

Attackers are shifting from traditional intrusion methods to credential theft, with increased use of infostealer malware and AI-enabled social engineering, leading to more incidents of attackers “logging in, not breaking in.”
Ransomware actors are moving away from Cobalt Strike in favour of native Windows tools due to declining ransom payment rates and a surge in data theft.
Akamai reported a convergence of Layer 7 DDoS, API abuse, and AI-powered attacks into highly coordinated, multi-vector campaigns that are more challenging to detect and mitigate.
Integrated Cloud Email Security (ICES) is highlighted as a recommended solution for modern email threats, as attackers continue to evolve phishing techniques.
SEO poisoning is being used to lure users searching for VPNs into downloading malicious software that steals corporate login credentials.
AI security remains a concern: studies indicate most CISOs are relying on outdated tools and skills to secure AI systems, and 90% of surveyed users do not trust AI with their data.
Use of in-memory loaders and open-source runtimes (such as Deno) is increasing in ransomware deployment for stealthier attacks.
Supply chain threats continue to target open-source repositories and extensions, as evidenced by the GlassWorm campaign.
Google is implementing stricter controls on Android apps abusing accessibility features to combat mobile malware.
A new tactic was observed where malicious commands are hidden from AI-based detection tools using font-rendering tricks in HTML.

Regulatory or Policy Developments

The European Union Council imposed sanctions on three entities and two individuals from China and Iran for involvement in cyberattacks targeting critical infrastructure.
Major tech companies including Anthropic, AWS, Google, Microsoft, and OpenAI have collectively invested $12.5 million in the Linux Foundation for open source security initiatives.
Google, Meta, and Microsoft are among the signatories of a new industry pact to combat online scams and fraud.

SGLang on DanSec Daily

Cybersecurity Brief – 2026-04-21

Major Incidents or Breaches

Newly Discovered Vulnerabilities

Notable Threat Actor Activity

Trends, Tools, or Tactics of Interest

Regulatory or Policy Developments

Cybersecurity Brief – 2026-03-18

Major Incidents or Breaches

Newly Discovered Vulnerabilities

Notable Threat Actor Activity

Trends, Tools, or Tactics of Interest

Regulatory or Policy Developments