Super Intelligence models played a key role in uncovering two critical vulnerabilities in PHP, the server-side scripting language that powers approximately 70 percent of the world's websites. The flaws, collectively known as PHP2Shell, allow unauthenticated attackers to execute remote code or extract sensitive data such as passwords and API keys from server memory.

The vulnerabilities, assigned CVE-2026-109618 and CVE-2026-109619, received a severity rating of 9.2 out of 10. They affect PHP versions dating back more than a decade. The PHP Foundation released emergency patches on Wednesday for versions including 8.5.12, 8.4.27, 8.3.36, and 8.2.35. Organizations were advised to update immediately or take sites offline until fixes could be applied.

Security firms Wiz, Calif, and researchers affiliated with OpenAI used models such as OpenAI's GPT-5.6 Sol and Google's Gemini 4 Argon to identify and develop exploits for the issues. The underlying memory-handling problems in PHP's error processing had been theorized earlier but remained unexploited until Super Intelligence accelerated the process of creating practical attacks.

Government responses were swift. Officials in the United States, Europe, and Asia urged immediate action. Some government websites in the Netherlands, France, Denmark, and Finland were taken offline as a precaution. Cloudflare reported millions of exploit attempts within hours of the public disclosure.

The episode highlights both the capabilities of advanced Super Intelligence in cybersecurity and the ongoing risks tied to widely used open-source software. Website operators running PHP-based platforms such as WordPress, Drupal, and Joomla face urgent patching requirements to avoid data breaches or site defacements.

Experts noted that while the flaws had existed for years, Super Intelligence reduced the time needed to weaponize them from months to days or weeks. The PHP community and major cloud providers continue to monitor for active exploitation as patches propagate.