Heap-based buffer overflow in Windows DNS Client allowing unauthenticated remote code execution with no user interaction. Exploitation requires attacker to intercept or respond to DNS requests through DNS spoofing, rogue DNS server, or man-in-the-middle position.
Apply May 2026 Patch Tuesday updates immediately. Use DNSSEC validation where possible. Implement network monitoring for DNS spoofing attempts. Use encrypted DNS (DoH/DoT) to prevent man-in-the-middle attacks.
Source: MSRC • Published: 2026-05-13
"NGINX Rift" heap buffer overflow in ngx_http_rewrite_module present since 2008 allowing unauthenticated remote code execution via single crafted HTTP request. In-the-wild exploitation began May 18-19, 2026. No authentication or user interaction required.
Update NGINX immediately to patched version. Implement WAF rules to detect and block crafted HTTP requests targeting rewrite module. Monitor for suspicious worker process behavior. Consider temporary mitigation by disabling rewrite module if not essential.
Source: Web • Published: 2026-05-13
"Dirty Frag" vulnerability (part 1 of 2-bug chain) enabling page-cache-to-root privilege escalation bypassing Copy Fail mitigations through xfrm-ESP or rxrpc. Full exploit published after embargo break with no patch or CVE initially assigned.
Monitor for kernel updates addressing Dirty Frag. Blacklist xfrm-ESP and rxrpc modules if not required. Implement enhanced privilege monitoring and detection. Review system logs for unauthorized privilege escalation attempts.
Source: Web • Published: 2026-05-07
"Dirty Frag" vulnerability (part 2 of 2-bug chain) enabling page-cache-to-root privilege escalation bypassing Copy Fail mitigations. Combined with CVE-2026-43284 to achieve local privilege escalation from unprivileged user to root.
Monitor for kernel updates addressing Dirty Frag chain. Implement defense-in-depth privilege controls. Use AppArmor or SELinux to restrict capabilities. Monitor for unusual local privilege escalation patterns.
Source: Web • Published: 2026-05-07
"Bleeding Llama" - out-of-bounds read flaw in Ollama allowing remote unauthenticated attacker to leak entire process memory. Impacts over 300,000 servers globally running the popular open-source LLM framework.
Update Ollama to patched version immediately. Restrict network access to Ollama servers using firewall rules. Implement memory protections and ASLR. Monitor for unusual memory access patterns. Rotate any credentials that may have been exposed.
Source: Web • Published: 2026-05-20
Mini Shai-Hulud campaign compromised TanStack npm packages affecting two OpenAI employee devices. Attacker published 84 malicious versions across 42 @tanstack/* packages on May 11, 2026. CI pipeline manipulated to steal its own publish token, leading to credential exfiltration from internal repositories.
Check package.json for TanStack packages installed on May 11, 2026 between 19:20-19:26 UTC. Rotate all credentials and certificates. Implement package integrity verification. Use package lock files and verify checksums. Monitor CI/CD pipelines for token theft.
Source: Web • Published: 2026-05-11
Massive supply chain attack where hackers compromised developer account and released over 630 malicious versions across 317 packages in 20 minutes. Malware steals credentials from password managers and spreads further. Includes compromise of Alibaba's Antv library.
Audit all dependencies installed on May 19, 2026. Use dependency scanning tools to identify compromised packages. Implement package signing verification. Rotate credentials stored in development environments. Enable 2FA for all package registry accounts.
Source: Web • Published: 2026-05-19
Comprehensive analysis uncovered 24 CVEs across popular AI IDEs with 100% vulnerable to prompt injection attacks. When combined with legacy IDE features, these enable remote code execution and data exfiltration from developer machines.
Update all AI-enhanced IDEs to latest versions. Disable AI features when working with untrusted code. Implement network monitoring for unusual data exfiltration. Use dedicated development environments isolated from sensitive data.
Source: Web • Published: 2026-05-20
"Rules File Backdoor" attack technique enables silent compromise of AI-generated code by injecting hidden malicious instructions using unicode characters and evasion techniques in model-facing instruction payloads. Attack bypasses typical code reviews and remains invisible to developers.
Implement strict code review processes for all AI-generated code. Use static analysis tools to detect hidden unicode characters. Disable custom rules files from untrusted sources. Monitor AI IDE configuration files for unauthorized modifications.
Source: Web • Published: 2026-05-20
"NomShub" vulnerability chain in Cursor AI enabling attackers to hijack developer machines via prompts hidden in malicious repositories. Exploits indirect prompt injection and command sandbox bypass to write code and abuse remote tunnel feature for shell access. Affects macOS systems where Cursor runs without sandbox restrictions.
Update Cursor AI immediately. Enable sandbox restrictions on macOS. Disable remote tunnel feature if not required. Review repository contents before opening in Cursor. Implement EDR solutions to detect unusual IDE behavior.
Source: Web • Published: 2026-05-20
Active supply chain attack targeting DAEMON Tools Windows installers since April 8, 2026. Installers from legitimate website signed with valid certificates but trojanized. Versions 12.5.0.2421 to 12.5.0.2434 confirmed compromised. Attack remains active as of May 20, 2026.
Uninstall DAEMON Tools versions 12.5.0.2421-12.5.0.2434 immediately. Perform full malware scan. Monitor for indicators of compromise. Await vendor notification of clean installer availability. Verify digital signature validity date before installation.
Source: Web • Published: 2026-04-08
JDownloader website compromised May 6-7, 2026 via unpatched CMS vulnerability. Alternative Windows and Linux installer download links altered to deliver malware: Python-based RAT for Windows, ELF binaries with root persistence for Linux. Main JAR package, in-app updates, macOS, Flatpak, Winget, and Snap unaffected.
Users who downloaded alternative installers May 6-7, 2026 should perform full system scan and reinstall from verified sources. Monitor for persistence mechanisms. Check for unauthorized scheduled tasks (Windows) or cron jobs (Linux). Rotate credentials.
Source: Web • Published: 2026-05-06
TeamPCP threat actor claimed responsibility for compromising hundreds of packages associated with major projects to push malware and steal developer credentials. Campaign includes CI pipeline manipulation to steal publish tokens. TeamPCP announced supply chain attack contest offering $1,000 in Monero.
Audit all dependencies from affected projects. Implement strict CI/CD pipeline security controls. Use isolated environments for token generation. Enable comprehensive audit logging. Report suspicious packages to registries immediately. Monitor dark web for credential leaks.
Source: Web • Published: 2026-05-20
PRC-linked Salt Typhoon successfully targeted U.S. House Committee staff emails, specifically focusing on congressional personnel working on national security committees with China oversight. Follows previous breaches of major U.S. telecommunications carriers by same threat actor.
Congressional staff should reset email passwords and enable MFA immediately. Review email access logs for unauthorized access. Implement zero-trust architecture for email systems. Conduct threat hunt for Salt Typhoon IOCs. Increase monitoring of national security personnel communications.
Source: Web • Published: 2026-01-09
ShinyHunters ransomware group claims theft of 275 million records from Instructure Canvas LMS affecting 8,809 educational institutions including students, teachers, and staff data. Cloud-hosted environment compromised with per-institution records ranging from tens of thousands to several million.
Canvas users should reset passwords immediately. Institutions should notify affected students and staff. Enable MFA on all Canvas accounts. Monitor for phishing attempts using stolen data. Review access logs for unauthorized access. Consider credit monitoring for affected individuals.
Source: Web • Published: 2026-05-20
Ransomware attack on West Pharmaceutical Services discovered May 4, 2026, resulting in data theft and encryption. Company is major manufacturer of drug-delivery devices and solutions. Critical systems including manufacturing, receiving, and shipping at certain locations were impacted.
West should continue forensic investigation and system restoration. Manufacturing partners should verify supply chain continuity. Healthcare organizations using West products should prepare for potential delivery delays. Monitor for stolen data publication on leak sites.
Source: Web • Published: 2026-05-04
ShinyHunters breached Vimeo through third-party vendor, accessing user and customer data including technical data, video titles, metadata, and customer email addresses. Part of broader ShinyHunters campaign targeting multiple organizations.
Vimeo users should reset passwords and enable 2FA immediately. Review account for unauthorized access or video modifications. Monitor email for phishing attempts. Update security questions. Consider changing email if used for password recovery.
Source: Web • Published: 2026-05-20
Medtronic experienced data breach on April 18, 2026, potentially exposing sensitive personal and health-related information of nearly 9 million individuals. Medical device manufacturer's services and systems were compromised. Full scope of breach under investigation.
Affected individuals should monitor for medical identity theft. Enable credit monitoring services. Review medical records for unauthorized access or changes. Report suspicious medical bills immediately. Change passwords for Medtronic patient portals. Monitor explanation of benefits statements.
Source: Web • Published: 2026-04-18
Home security giant ADT suffered breach impacting 5.5 million users after attackers accessed Salesforce cloud through compromised Okta SSO login. Third breach for ADT since 2024. Alarm systems and payment data reportedly unaffected but customer data compromised.
ADT customers should reset account passwords and enable MFA. Review account access logs for unauthorized changes. Update security questions. Monitor for social engineering attempts using stolen data. Consider additional home security monitoring given breach of security provider.
Source: Web • Published: 2026-05-20
Russian state-sponsored APT28 (Fancy Bear/Forest Blizzard) exploiting CVE-2026-21510, CVE-2026-21513, and CVE-2026-32202 since December 2025 targeting Ukraine and EU organizations. Attack chain involves Windows Shell RCE, malicious LNK files, MSHTML security feature bypass, and zero-click NTLM authentication coercion.
Apply patches for CVE-2026-21510, CVE-2026-21513, and CVE-2026-32202 immediately. Block NTLM authentication where possible. Implement strict LNK file handling policies. Deploy EDR solutions to detect multi-stage attack chains. Monitor for Russian TTPs and IOCs shared by CISA.
Source: GTIG • Published: 2026-05-20
China-nexus APT group UAT-8302 targeting government entities in South America since late 2024 and southeastern Europe in 2025. Deploys multiple custom malware families including .NET-based "NetDraft" backdoor (C# variant of FinalDraft/SquidDoor). Sophisticated post-compromise operations with malware previously used by other known China-nexus actors.
Government entities should conduct threat hunt for UAT-8302 IOCs. Implement enhanced monitoring for NetDraft/FinalDraft backdoor indicators. Review network access logs for unusual C2 communications. Deploy EDR solutions capable of detecting custom malware. Coordinate with national CERT teams for intelligence sharing.
Source: Web • Published: 2026-05-20
Ransomware operators in 2026 increasingly prioritize neutralizing endpoint defenses before payload execution. EDR killers now standard component of attack playbooks. Attackers exploit trusted signed drivers through BYOVD technique to terminate security processes and disable monitoring, blending into legitimate system activity while degrading defensive visibility.
Implement driver signature verification and blocklists for known vulnerable drivers. Enable protected processes for security software. Deploy tamper protection features. Monitor for security service termination attempts. Use multiple defensive layers beyond EDR. Implement application control to prevent BYOVD exploitation.
Source: Web • Published: 2026-05-20
Digital training platform Udemy identified as victim in broad ShinyHunters ransomware campaign. Breach details and scope of data compromised under investigation. Part of multi-organization attack affecting educational and training platforms.
Udemy users should reset passwords and enable MFA. Monitor accounts for unauthorized purchases or course enrollments. Review account activity logs. Be vigilant for phishing emails claiming to be from Udemy. Update payment methods if stored on platform.
Source: Web • Published: 2026-05-20
Spanish fast-fashion retailer Zara database breach exposing data of more than 197,000 customers. Compromised information includes customer personal details associated with online shopping accounts.
Zara customers should reset passwords immediately. Enable account notifications for purchases. Monitor payment methods for fraudulent transactions. Be alert for phishing attempts. Consider changing passwords on other sites if reused. Review account purchase history for unauthorized orders.
Source: Web • Published: 2026-05-20
NVIDIA GeForce NOW Alliance partner in Armenia breached by threat actor claiming ShinyHunters affiliation. User database exposed including names, emails, nicknames, dates of birth, membership details, 2FA status, and internal roles.
GeForce NOW Armenia users should reset passwords immediately. Enable or reset 2FA. Monitor accounts for unauthorized access. Be alert for targeted phishing using stolen personal information. Review account activity for suspicious sessions. Update recovery email if compromised.
Source: Web • Published: 2026-05-20
Google Threat Intelligence Group reports first confirmed case of AI model discovering and exploiting zero-day vulnerability for 2FA bypass. Criminal threat actor planned mass exploitation but Google's proactive discovery may have prevented deployment. GTIG has high confidence this represents escalation in AI-assisted vulnerability discovery.
Organizations should accelerate vulnerability patching cycles assuming AI-accelerated discovery. Implement defense-in-depth for authentication beyond 2FA. Deploy behavioral analytics to detect novel exploitation patterns. Monitor for rapid zero-day weaponization. Coordinate with threat intelligence providers for AI-discovered vulnerability IOCs.
Source: GTIG • Published: 2026-05-20
Prompt injection remains OWASP LLM01 top security risk in April 2026. Multi-turn jailbreaks now preferred attack vector on frontier models. Multimodal injections (images, QR codes, steganographic payloads) matured. MCP server exploitation emerged as new attack surface. Research shows 73% of production AI deployments vulnerable to prompt injection.
Implement input validation and sanitization for all LLM prompts. Use prompt guards and content filters. Deploy multimodal input analysis. Restrict LLM tool access and capabilities. Implement output validation. Monitor for jailbreak attempts. Apply principle of least privilege to MCP servers.
Source: OWASP • Published: 2026-04-20
Shadow AI now third most common non-malicious insider action in DLP datasets in 2025, representing fourfold increase from previous year. LLMs helping attackers create exploits, execute documented techniques at scale, and improve phishing campaigns. Represents growing insider risk and attacker enablement.
Implement AI usage policies and approved tool lists. Deploy DLP solutions with AI/LLM detection capabilities. Provide security training on AI risks. Monitor for data exfiltration to unauthorized AI services. Implement network controls to restrict unapproved AI tool access. Create sanctioned AI environments for legitimate use.
Source: Web • Published: 2026-05-20
CVE-2026-41096: Heap-based buffer overflow in Windows DNS Client allowing unauthenticated remote code execution with no user interaction. Exploitation requires attacker to intercept or respond to DNS requests through DNS spoofing, rogue DNS server, or man-in-the-middle position.