ARTFEED — Contemporary Art Intelligence

PHP Patches Two Critical Remote Code Execution Flaws Found With AI Assistance

digital · 2026-10-21

PHP issued security updates on Wednesday addressing two critical vulnerabilities that permit unauthenticated remote code execution. Researchers located both flaws with assistance from AI models. The PHP security team had set the release for 14:00 UTC, which corresponds to 10 a.m. ET. Each vulnerability received a severity rating of 9.2 on the CVSS v4.0 scale. The flaws allow attackers to execute code remotely without authentication, a class of vulnerability considered among the most dangerous for server-side software. AI models contributed to the discovery process, a detail that places the release within a growing pattern of machine-assisted security research. The patches were made available through official PHP security updates. No further technical details about the individual vulnerabilities were provided in the source material.

Key facts

  • PHP released security updates on Wednesday
  • The updates fix two critical vulnerabilities
  • Both flaws allow unauthenticated remote code execution
  • Researchers found the flaws with help from AI models
  • The release was scheduled for 14:00 UTC (10 a.m. ET)
  • Each flaw was rated 9.2 on the CVSS v4.0 scale
  • The PHP security team handled the release
  • AI models assisted in the discovery of both vulnerabilities

Entities

Artists

  • Lukasz Olejnik
  • Davi Ottenheimer
  • Brett Leatherman

Institutions

  • PHP
  • PHP Foundation
  • Wiz
  • Google
  • OpenAI
  • Calif
  • Cloudflare
  • Sansec
  • King's College London
  • National Land Survey of Finland
  • Port of Rotterdam
  • EU-CERT
  • National Cyber Security Center
  • FBI
  • Bleeping Computer
  • Ars Technica
  • Reuters
  • WordPress
  • Drupal
  • Adobe
  • Magento

Locations

  • Finland
  • France
  • Rotterdam
  • Netherlands
  • United Kingdom
  • Australia
  • United States

Sources