AI Models Turn Dormant PHP Flaws Into Critical Web Threats

The PHP Foundation released emergency patches for its core programming language on Wednesday, October 21, moving the update forward from its scheduled release date. This accelerated rollout follows the discovery that artificial intelligence models turned a long-standing code defect into a high-risk security threat that could impact a vast majority of websites online. The coordinated fix addresses two critical vulnerabilities that had eluded human developers for two years but were weaponized by AI-assisted researchers in a matter of weeks.

The primary driver behind this urgent response was the demonstration that GPT-5.6 Sol, an AI model from OpenAI, could identify and exploit specific memory management errors in PHP. These flaws, tracked as CVE-2026-109618 and CVE-2026-109619, allow attackers to remotely execute malicious code on servers without needing login credentials. The vulnerabilities arise when the PHP engine mishandles allocated memory during high-load requests or fatal errors, potentially leaving behind incomplete data or pointing to freed memory that can be corrupted by subsequent inputs.

Both defects carry a critical severity score of 9.2 out of 10, reflecting their potential to expose sensitive server data such as passwords, authentication tokens, and API keys. Because these issues reside in the core PHP engine rather than in individual application code, virtually any website built on the language is susceptible. The PHP Foundation has issued fixed versions 8.2.35, 8.3.36, 8.4.27, and 8.5.12, while versions 8.1 and older will not receive official updates, requiring users to upgrade to newer branches to remain secure.

The discovery process began on July 30, 2026, when Calif, a cybersecurity firm and founding partner of OpenAI’s “Patch the Planet” initiative, started auditing PHP at the request of the PHP Foundation. Although the Foundation had been aware of the underlying flaw since 2024, human teams could not find a way to exploit it or fix it. On August 11, Calif researcher Ryan A. identified the exploitable path, with OpenAI researchers confirming the vulnerability the same day. Separately, the firm Wiz independently found a second, related flaw, leading to the joint disclosure.

OpenAI attributes the two-month timeline for the fix to the complexity of altering the PHP engine’s fundamental operations. The company frames this incident as a real-world example of its “Defender’s Window” concept, a narrow period where organizations can use frontier AI models to patch defenses before adversaries do. OpenAI argues that since powerful AI tools will inevitably be used by threat actors, it is a collective responsibility to proactively secure critical infrastructure. The company noted that while human experts struggled with the code for two years, AI-wielding researchers developed working exploits in just two weeks.

The premature disclosure on October 20, which sparked public concern, forced the Foundation to pull the release forward from October 22. OpenAI considers this one of the most significant cybersecurity events of 2026 due to PHP’s estimated presence on 70% of the internet. The incident highlights a broader trend in which AI models are uncovering deep-seated bugs in major software ecosystems, such as the thousands of zero-day vulnerabilities recently identified in operating systems and browsers by Anthropic’s Claude Mythos.

For site administrators in Texas and across the South Central region, the immediate action required is to verify their PHP versions. If a site is running version 8.2 or higher, it should be updated to the latest patched release immediately. Those on version 8.1 or older must plan for a more significant upgrade, as no official patches are available for those legacy versions. This event underscores the shifting landscape of web security, where AI is no longer just a tool for building applications but a primary force in both breaking and securing them.