PHP2Shell Attacks Escalate with AI Automation, Lazarus Group Suspected
Original source
PHP2Shell Exploitation Hits Government Entity Amid Global Scanning and Signs of AI-Assisted Hacking
The Hacker News →Two critical PHP vulnerabilities (CVE-2026-109618 and CVE-2026-109619) are under active exploitation, with attackers deploying agentic AI systems for automated reconnaissance following initial compromise. A documented attack against a European government entity achieved access to password hashes, MFA secrets, and an NTLM-relay tunnel to Active Directory within two hours, with AI-generated commands formatted for machine-readable output to enumerate systems, hunt privileged accounts, and escalate permissions.
Parallel reconnaissance activity reveals a suspected North Korean campaign coordinating across 21 compromised nodes to scan PHP infrastructure. The operation targets 216 unpatched or partially mitigated systems spanning government, financial institutions, aviation, logistics, manufacturing, and transaction platforms across multiple countries. Korean-speaking operators using I2P channels employ deliberate low-and-slow scanning methods to evade detection thresholds and rotate nodes at the first sign of blocking. Activity has progressed from reconnaissance to planning active exploitation of prioritized targets.
Cloudflare has observed millions of exploitation attempts since public disclosure. Security researchers identified specific error sequences in PHP and Apache logs that signal exploitation. A third RCE vulnerability (CVE-2026-115560) exists under uncommon configuration scenarios. Bitsight scans indicate nearly 37 million internet-facing systems run PHP or related software, primarily in the U.S., Germany, France, the Netherlands, and Japan. PHP runtime vulnerabilities require direct host updates rather than application patches or WAF mitigation.
Read the full article
Continue reading at The Hacker News →This is an AI-generated summary. Read the original for the full story.