Search: CVE-2026-109619

PHP has various global state that temporarily points into request memory. In some scenarios, global state may not be correctly reset, or may be repopulated after being reset. The global state may then refer to request memory even after it has been freed. Subsequent requests may read or branch on this stale state before it is fully initialized. Multiple strategies can trigger this condition and produce remote code execution (RCE).

Amazon Threat Intelligence detected exploitation attempts of the PHP2Shell vulnerability in MadPot honeypot telemetry on October 17, 2026, four days before the public disclosure of CVE-2026-109618 and CVE-2026-109619. The initial exploit originated from a commercial anonymization network, indicating threat actor activity during the embargo period. Multiple proof-of-concept exploits appeared on the disclosure date with diverse payloads and infrastructure, suggesting involvement of multiple actors. After the embargo lifted on October 21, 2026, exploitation rapidly became widespread. This incident demonstrates that coordinated vulnerability disclosure does not guarantee zero-day protection and that AI-assisted exploit development has significantly reduced the time from patch release to active exploitation.

Two critical remote code execution vulnerabilities in PHP, collectively named PHP2Shell, were disclosed on October 21, 2026. CVE-2026-109618 involves out-of-memory handling corruption that can interrupt PHP while internal request data is partially initialized, leading to invalid data treatment. CVE-2026-109619 exploits stale global state where request-owned internal state survives after memory is freed. Both vulnerabilities can be exploited successfully on ARM64 systems and x86 systems without ASLR. Fixed versions include PHP 8.5.12, 8.4.27, 8.3.36, and 8.2.35. Over 2.6 million internet-facing hosts running PHP were identified as vulnerable, with 42% on end-of-life branches receiving no official patches. Early exploitation attempts were observed from multiple sources, including scanning activity and webshell deployment attempts targeting the upload-progress vulnerability path.

EPSS 11%Campaign#mirai botnet#woocommerceUnited States+2

Researchers uncovered an AI-driven intrusion ecosystem operated by affiliates of INC and The Gentlemen ransomware groups via seven exposed command-and-control servers. The operators integrated commercial AI models, exploited edge devices at scale, and compromised thousands of targets, primarily in the US industrial manufacturing and healthcare sectors. The infrastructure included multiple custom encryptor builds and evidence of active victim encryption. Operational security failures exposed source code, credentials, AI session logs, and victim data. The campaign demonstrates advanced ransomware-as-a-service operations leveraging AI and custom tooling.

Malware#simplehelp#phantom frameworkUnited States

A financially motivated threat actor named BUGSTER used jailbroken AI assistants, including the Hermes Agent framework with DeepSeek V4, to conduct multiple cybercrime activities primarily in India and the British Indian Ocean Territory. The actor exploited AI-generated scripts to exfiltrate over 5,200 personal records from a university database, perform WordPress brute force attacks, and plan VoIP telecommunications fraud. Additional malicious activities involved Wi-Fi password cracking requiring physical proximity, cryptocurrency whale reconnaissance, and credential scanning toolkit development. This campaign demonstrates how AI tools can lower the skill barrier for cybercriminal operations.

Campaign#ai-enabled attacks#deepseekBritish Indian Ocean Territory+1

A malware campaign used a sponsored Google search result to redirect victims to a spoofed Amazon login page on Google Sites. The page displayed a fake CAPTCHA (ClickFix) that tricked users into running a malicious command via the Windows finger utility, initiating a multi-stage infection. Payloads included IronPython, obfuscated Python scripts, Doina malware, Lumma Stealer via a Node.js package, and a pirated NetSupport Client RMM tool. The campaign used barbecue-themed domains for command and control and had alternate payloads for macOS. Attackers employed advanced obfuscation techniques such as Base64 with Cyrillic substitution, XOR, and RC4 encryption.