PHP Patches Fix Two-Year-Old AI-Found Flaw

The core tension in this week's cybersecurity headlines is the speed gap between human developers and artificial intelligence. A critical flaw in PHP, the open-source language that powers an estimated 70% of the internet, sat undetected in terms of exploitability for two years. Human teams knew about the underlying code issue as early as 2024 but could not figure out how to break it. AI-assisted researchers, however, turned that dormant bug into a working exploit in just two weeks, forcing a rapid, coordinated patch release to protect a vast portion of the web.

OpenAI announced on October 21 that its GPT-5.6 Sol model was the engine behind the discovery. The work was conducted under the 'Patch the Planet' initiative, a cybersecurity program involving the California-based firm Calif, the PHP Foundation, and other partners. The two vulnerabilities, identified as CVE-2026-109618 and CVE-2026-109619, both carry a critical severity score of 9.2 out of 10. The flaws are not in the application code but in PHP's core engine, meaning even perfectly written websites could be compromised if they run an unpatched version of the language.

The technical specifics involve memory management errors. The first vulnerability occurs when PHP runs out of allocated memory during a request, potentially leaving incomplete internal data that the engine later treats as valid. The second flaw happens when a fatal error causes the engine to point to memory that has already been freed. This allows subsequent requests to access or corrupt that space. Because these bugs exist in the core, attackers can exploit them remotely without needing login credentials, potentially executing malicious code and taking control of servers.

The timeline of the discovery was tightly managed. Researchers from Calif began auditing PHP on July 30, 2026, at the request of the PHP Foundation. On August 11, Calif researcher Ryan A. identified the first exploitable vulnerability, with OpenAI researchers confirming a similar finding that same day. The cybersecurity firm Wiz independently identified the second vulnerability shortly after. The coalition then spent two months working on a fix, a duration OpenAI attributed to the complexity of making major changes to the PHP engine.

The release schedule was thrown into chaos by premature disclosures on October 20. Originally planned for October 22, the patched versions were released on October 21 to address public outcry. The PHP Foundation has now released updates for versions 8.2.35, 8.3.36, 8.4.27, and 8.5.12. Crucially, PHP 8.1 and older versions will not receive official fixes, leaving those installations vulnerable unless users upgrade to a supported version.

OpenAI is using this incident to champion its 'Defender's Window' concept. This theory suggests there is a narrow timeframe where organizations can use frontier AI models to find and patch defenses before malicious actors gain access to the same capabilities. The company argues that because powerful technology will inevitably be used by bad actors, it is a collective responsibility to act proactively. This view is supported by recent reports that Anthropic's Claude Mythos model has identified thousands of zero-day vulnerabilities in major operating systems and browsers.

The incident has sparked debate about the nature of AI-driven security risks. Some argue that if humans could not exploit the flaw for two years, it was effectively harmless until AI arrived. OpenAI challenges this, stating that the existence of the flaw meant a risk was always present, and the AI merely accelerated the exploitation. The company considers this one of the biggest cybersecurity discoveries of 2026 due to the sheer ubiquity of PHP.

For the millions of websites running on the language, the immediate stakes are high. Any site using PHP must upgrade to the latest patched version to remain secure. The ability of the AI to expose sensitive information stored in server memory—including passwords, authentication tokens, and API keys—adds urgency to the update. The flaw's location in the core engine means that even secure application code cannot protect a server running an outdated PHP version.

The partnership between the PHP Foundation, Calif, and OpenAI highlights a new model for vulnerability management. By leveraging AI tools, researchers moved from a theoretical bug to a practical exploit in a fraction of the time it would have taken traditional methods. This speed, while impressive, also compresses the window for developers to react. The October 21 release serves as a wake-up call for the broader tech industry, suggesting that the pace of AI discovery will continue to outstrip traditional human-driven security audits.