Find what you’re looking for
Browse everything
6,838 results in Threat Reports
Silent Patch, Critical Risk: How a Malicious MCP Server Can Own a Developer's Machine Through OpenCode
In late September 2026, a critical remote code execution vulnerability (severity 9.8/10) was discovered in OpenCode, an AI coding agent developed by Anomaly. The flaw allowed a malicious or compromised MCP (Model Context Protocol) server to execute arbitrary code on a developer's Windows or WSL machine during the OAuth sign-in flow. OpenCode passed the server-supplied authorization endpoint URL to a PowerShell-based browser-launch helper without sanitizing it, enabling command injection. Affecting versions 1.0.137 through 1.18.32, the bug was silently patched in version 1.18.33 on September 28, 2026, without a CVE assignment or public advisory, leaving many installations unknowingly exposed. This incident underscores the expanding attack surface introduced by AI agent ecosystems and MCP server integrations. As agentic AI tooling proliferates in developer workflows, supply-chain and third-party plugin compromises represent a growing vector. The absence of a CVE means automated scanners cannot flag affected versions, amplifying organizational risk and highlighting gaps in AI-era software disclosure practices.
PHP2Shell: Critical Unauthenticated RCE Hits Every Major PHP Server API
In October 2026, the PHP project disclosed two critical unauthenticated remote code execution vulnerabilities, CVE-2026-109618 and CVE-2026-109619, collectively dubbed PHP2Shell, affecting all mainstream PHP server APIs including PHP-FPM, Apache mod_php, php-cgi, FrankenPHP, and LiteSpeed. Both flaws carry a CVSS 4.0 score of 9.2 and require no user interaction or prior authentication. CVE-2026-109618 exploits out-of-memory corruption during partial request initialization, while CVE-2026-109619 leverages stale global state surviving across worker requests. Fixed versions 8.5.12, 8.4.27, 8.3.36, and 8.2.35 were released October 21, 2026, with wild exploitation confirmed on the same day. PHP2Shell represents the most broadly impactful PHP vulnerability class in years, exposing virtually every organization running PHP web infrastructure. With exploitation confirmed in the wild and no viable mitigation short of patching, this incident underscores the accelerating trend of unauthenticated, pre-authentication RCE attacks targeting foundational web runtime components rather than application-layer logic.
How Storm-0940 Used AI to Hack Hundreds of Companies in Hours via ConnectWise
Beginning September 2026, Storm-0940, a China-state aligned threat actor active since 2021, weaponized ARTEX—an open-source agentic AI penetration-testing framework—to automate large-scale attacks against hundreds of organizations across aerospace, defense, financial services, government, and technology sectors globally. The actor exploited a supply chain compromise of ConnectWise's multi-tenant cloud backup application, using ARTEX to conduct password spraying at 2,900 attempts per minute, OAuth token theft, and Microsoft Graph reconnaissance. In several confirmed intrusions, privileged cloud access was achieved in under three hours, with follow-on activity including application credential modification, privileged role self-assignment, and persistent backdoor establishment across victim Microsoft 365 tenants. This incident marks a watershed moment in nation-state offensive operations: AI-driven automation has collapsed the attacker dwell-time advantage that defenders historically relied upon. As agentic AI frameworks become commoditized and open-source, the barrier to executing sophisticated, multi-stage supply chain compromises at machine speed is dramatically lowered, forcing organizations to rethink detection and response timelines and elevate cloud identity governance to a board-level priority.
Two Critical PHP RCE Vulnerabilities Patched in 2026: What Security Teams Must Do Now
In October 2026, PHP maintainers disclosed and patched two critical unauthenticated remote code execution vulnerabilities tracked as CVE-2026-109618 and CVE-2026-109619. The first flaw exploits out-of-memory handling that interrupts PHP during partial request initialization, causing request cleanup to treat invalid data as valid and enabling arbitrary code execution. The second flaw allows request-owned internal state to persist after its backing memory is freed, letting a subsequent request handled by the same worker consume stale state before initialization resets it. All major PHP server configurations—including mod_php, PHP-FPM, php-cgi, FrankenPHP, and LiteSpeed—are affected, with mod_php carrying the highest demonstrated risk. No authentication is required to trigger either vulnerability, meaning exploitation can occur before any application-layer code executes. These vulnerabilities represent a systemic risk to the global web stack, as PHP powers an estimated 77% of server-side websites. The disclosure arrives amid a broader surge in exploitation of web-tier vulnerabilities by both opportunistic actors and nation-state groups, reinforcing the urgency of rapid patching cycles and layered network controls to contain blast radius when patch deployment is delayed.
Storm-0940 and ARTEX: How China's AI-Powered Agents Hijacked ConnectWise and 2,250 Microsoft 365 Tenants
In October 2026, Microsoft attributed a sophisticated supply-chain attack against ConnectWise's Cloud Backup service to Storm-0940, a China-state aligned threat actor active since 2021. The group leveraged ARTEX, an open-source agentic AI penetration testing framework, to automate password-spraying at 2,900 sign-in attempts per minute, ultimately targeting more than 2,250 Microsoft 365 tenants across eight organizations. AI agents autonomously progressed from initial authentication through reconnaissance, OAuth abuse, credential modification, and privilege escalation with minimal human interaction. ConnectWise temporarily shut down services following the attack, and Microsoft confirmed evidence of follow-on persistence activity in multiple customer environments. This incident marks a critical inflection point in nation-state tradecraft: the weaponization of agentic AI frameworks enables adversaries to compress attack timelines from days to minutes, outpacing human defenders. The combination of supply-chain entry, AI-driven automation, and OAuth abuse signals a new era of high-velocity, low-visibility intrusions that demand equally automated defensive countermeasures.
When the Number 4 Kills a Certificate Authority: The eMudhra PKI Collapse
In April 2025, Indian certificate authority eMudhra conducted a key generation ceremony that produced critically weak private keys for eight intermediate SSL/TLS certificates. The keys were so poorly generated — some as small as the integer 4 — that security researcher Scott Helme reconstructed all eight on a laptop in under one second. Major browser vendors including Chrome, Firefox, Apple, and Microsoft revoked trust within days. Mozilla removed eMudhra's root certificate entirely, and Google announced imminent Chrome root removal. Facing regulatory scrutiny from CERT-In, a 40% share price collapse, and reputational damage, eMudhra's board voted to exit the certificate authority business altogether. This incident highlights how a single procedural failure in a cryptographic ceremony can instantaneously nullify an entire trust hierarchy. As PKI underpins HTTPS, code signing, and enterprise identity, the fragility exposed here is resonating across the industry at a moment when regulators and browser consortiums are tightening CA accountability standards and scrutinizing key generation practices globally.
PHP2Shell (CVE-2026-109618 & CVE-2026-109619): How to Use AWS WAF as a Bridge While You Patch
On October 21, 2026, the PHP project issued emergency patches for two critical unauthenticated remote code execution vulnerabilities—CVE-2026-109618 and CVE-2026-109619—collectively named PHP2Shell, both rated 9.2 (Critical) under CVSS v4.0. The flaws reside in PHP's multipart/form-data request processing layer, meaning attackers can trigger arbitrary code execution before any application logic runs. CVE-2026-109618 exploits an out-of-memory memory corruption condition, while CVE-2026-109619 abuses stale global state persisting across PHP worker processes. Fixed releases are PHP 8.5.12, 8.4.27, 8.3.36, and 8.2.35. Amazon Threat Intelligence detected active exploitation via MadPot prior to public disclosure, confirming real-world weaponization. PHP2Shell underscores how pre-application-layer vulnerabilities in widely deployed runtimes can bypass traditional application-level defenses entirely. With PHP powering an estimated 77% of server-side web infrastructure, the blast radius is enormous. Organizations relying solely on application-layer WAF rules without addressing runtime configuration and patch cadence remain exposed, highlighting the necessity of layered, defense-in-depth network security architectures that extend from the edge to workload internals.
When the Lock Itself Is Broken: The eMudhra PKI Compromise and What It Means for Encrypted Infrastructure
In October 2026, certificate authority eMudhra disclosed that eight intermediate CA certificates issued in April 2025 contained critically weak P-384 private keys with values as low as 4—making them trivially recoverable by any attacker. Security researcher Wayne discovered the flaw on October 17, 2026, while scanning public certificate transparency logs. Because major browsers and operating systems trusted these certificates, a malicious actor possessing the recovered private keys could have forged trusted TLS certificates for any website, enabling man-in-the-middle interception of encrypted communications at scale. Mozilla, Google Chrome, and Apple responded within days by revoking and blocking the affected intermediates, and eMudhra halted issuance under its G3 root. This incident underscores a growing pattern of PKI supply-chain failures where procedural errors in certificate generation—rather than external exploits—introduce systemic trust vulnerabilities. As organizations increasingly depend on TLS for zero-trust enforcement and encrypted data-in-transit assurance, a single compromised CA intermediate can silently undermine those controls across entire ecosystems for months before detection.
SafePay Ransomware Has Breached 600+ Critical Infrastructure Organizations: What Security Teams Must Do Now
Since August 2024, the SafePay ransomware group has breached more than 600 critical infrastructure organizations across North America, South America, Europe, and Australia. Operating as a private, affiliate-free group, SafePay gains initial access through credential exposure, brute-force attacks, and exploitation of VPN edge devices—particularly leveraging Mullvad VPN IP addresses and phished account profiles. Once inside, actors perform double-extortion: exfiltrating sensitive data before encrypting systems, then threatening public release if ransoms go unpaid. Recovery is further undermined by deliberate deletion of shadow copies and wiping of network-attached storage backups. A joint advisory from CISA, FBI, USSS, and international partners was published October 21, 2026. SafePay has rapidly filled the vacuum left by dismantled ransomware operations such as BlackCat (ALPHV) and LockBit, emerging as one of the most active ransomware threats globally. With valid credentials bypassing prevention controls in 63% of post-access scenarios, the incident underscores escalating risks tied to identity abuse, VPN exploitation, and insufficient lateral movement controls across hybrid and cloud environments.
ClickFix in 2026: How Fake AI Tool Installers Became the Most Dangerous Initial-Access Technique of the Year
Throughout 2026, the ClickFix social-engineering framework has surged to unprecedented scale, accounting for 52% of browser-based attack detections through Q2 2026 and reaching 67% in August. Attackers now weaponize the workforce's trust in AI tools—cloning install pages for Claude Code, ChatGPT, and Codex (InstallFix), and planting malicious instructions inside legitimate AI-platform sharing features such as custom GPTs and shared Artifacts (LLMshare). Victims are funneled through search-engine malvertising and SEO poisoning, bypassing email gateways entirely. Once a user pastes the attacker-supplied terminal command, fileless malware executes in-memory, evading endpoint detection. Microsoft recorded ClickFix-style commands on more than 1.1 million unique devices between February and May 2026 alone, with 96.3% of intrusions resulting in malware deployment. ClickFix has evolved from a niche clipboard-hijack trick into the dominant initial-access technique of 2026, displacing AiTM and device-code phishing. Its convergence with AI-tool adoption habits—terminal installs and search-engine discovery—creates a self-reinforcing lure that is structurally difficult for traditional email, URL-reputation, and endpoint controls to stop, elevating it to a board-level supply-chain and insider-risk concern.
PHP2Shell: Emergency Patches for Critical Unauthenticated RCE Threatening 5 Million+ Web Servers
In October 2026, the PHP security team issued emergency patches for two critical unauthenticated remote code execution vulnerabilities—CVE-2026-109618 and CVE-2026-109619—collectively tracked as PHP2Shell. Discovered jointly by Calif, Wiz, and OpenAI researchers, the flaws affect all PHP web-serving configurations including PHP-FPM, Apache mod_php, php-cgi, FrankenPHP, and LiteSpeed. Both vulnerabilities exploit abnormal request termination leading to memory corruption and heap use-after-free conditions, enabling attackers to execute arbitrary code as the web worker before any application code runs. With over 5.1 million PHP hosts exposed online and early exploitation already observed by Cloudflare, the attack surface is enormous—impacting an estimated 75–80% of all websites globally. PHP2Shell represents a watershed moment for server-side web security: unauthenticated pre-application RCE at scale, affecting CMS platforms like WordPress, Joomla, and Drupal. The incident underscores rising attacker sophistication in targeting foundational runtime environments, a trend accelerating alongside the expansion of cloud-hosted PHP workloads and automated exploit tooling.
Plex2Shell: How Attackers Chained Four CVEs Into a 9.8 Critical RCE—and What It Means for Your Cloud Security Posture
In October 2026, CISA issued an urgent advisory warning that threat actors were actively chaining four vulnerabilities in Plex Media Server—collectively dubbed Plex2Shell—to achieve unauthenticated remote code execution on exposed servers. The exploit chain, assigned a critical CVSS score of 9.8, begins with an unauthenticated SSRF flaw (CVE-2026-96652) that allows attackers to reach internal APIs, progresses through a URL injection flaw (CVE-2026-96654) to invoke plugin functions, escalates via an authenticated SSRF (CVE-2026-96655), and culminates in arbitrary file write (CVE-2026-96656) to plant malicious shared libraries. Active exploitation was observed as early as October 17, 2026, initially targeting honeypots to deploy cryptominers. Federal agencies were directed to patch or mitigate by October 23, 2026. Plex2Shell illustrates the accelerating trend of researchers publishing full exploit chains and PoC code within days of responsible disclosure, dramatically compressing the window between public knowledge and mass exploitation. With over 36,000 Plex servers estimated to be internet-exposed, chained CVE attacks against consumer-grade media infrastructure highlight how attacker tooling now commoditizes multi-stage exploits for rapid, widespread campaigns.