70% of the Web at Risk: PHP Patches Released After AI Finds Critical Flaws

The PHP Foundation pulled forward its patch release to October 21 to address two critical security vulnerabilities that could affect an estimated 70% of all websites. This decision came after premature disclosures on October 20 sparked public concern, prompting the organization to move up the coordinated release from its original October 22 schedule. The fixes are included in versions PHP 8.2.35, 8.3.36, 8.4.27, and 8.5.12, while versions 8.1 and older will not receive official updates.

OpenAI announced the details in a press release on October 21, stating that researchers working under its "Patch the Planet" cybersecurity initiative used its GPT-5.6 Sol model to uncover the flaws. The company notes that it coordinated a coalition of researchers, PHP maintainers, security providers, and government cybersecurity agencies to resolve the issues over a two-month period. OpenAI attributed the lengthy timeline to the complexity of the task, which required significant changes to the core PHP engine.

The investigation began on July 30, 2026, when researchers from Calif, a California-based cybersecurity firm and founding partner of the initiative, started auditing PHP at the request of the PHP Foundation. The Foundation had been aware of the underlying flaw since 2024 but was unable to identify a method to exploit it or to fix it. Calif researcher Ryan A. identified the exploitable vulnerability on August 11, and OpenAI researchers did the same that day. Shortly after, another cybersecurity firm, Wiz, independently identified a second vulnerability.

The two issues, tracked as CVE-2026-109618 and CVE-2026-109619, stem from flaws in how PHP manages memory and handles errors. The first occurs when PHP runs out of allocated memory while processing a request, potentially leaving behind incomplete internal data that the engine later mistakes for valid information. The second involves a fatal error that leaves PHP pointing to memory that has already been freed, allowing subsequent requests to access or corrupt that space.

Under specific conditions, attackers can exploit these weaknesses remotely without needing login credentials. This allows them to execute malicious code on affected servers, potentially taking control of websites. Using AI tools, researchers also demonstrated that these flaws could expose sensitive data stored in server memory, including passwords, authentication tokens, and API keys. Both vulnerabilities received a critical severity score of 9.2 out of 10.

A significant concern is that these bugs exist in PHP's core engine. This means websites can be affected even if their application code is perfectly secure. OpenAI highlights this incident as one of the major cybersecurity discoveries of 2026 due to PHP's widespread use. The company cites this event as an example of its "Defender's Window" concept, a timeframe during which organizations can use advanced AI models to patch defenses before those same capabilities become widely available to malicious actors.

The situation raises questions about the role of AI in security. Human developers and security teams examined the underlying code for two years but could not fully resolve the flaws. In contrast, AI-wielding researchers demonstrated practical exploitation methods within two weeks. Some argue the flaws might not have been exploitable without AI, rendering them harmless. However, OpenAI challenges this view, arguing that powerful technology will inevitably be used by bad actors, making proactive defense a collective responsibility. Anthropic's Claude Mythos recently identified thousands of zero-day vulnerabilities in major operating systems and web browsers.

OpenAI states that every site using PHP can now safely upgrade to stay secure. The company introduced the Defender's Window concept after its models hacked Hugging Face, describing a critical, narrow period for strengthening cybersecurity defenses.