Image: PHP

OpenAI says its AI models helped find two critical security flaws in PHP, the programming language behind a large share of the web. PHP released fixes today, and OpenAI described its role on its blog and on X. OpenAI estimates PHP runs about 70% of websites and calls this one of the most serious security issues in recent memory.

OpenAI models helped researchers—operating as part of our Patch the Planet initiative—discover critical security flaws in PHP, an open-source programming language used to run an estimated 70% of websites.

Over the past two months, we led a coalition of maintainers, researchers, and security providers to prepare fixes and protections ahead of disclosure. This duration was necessary because the fix was quite intricate and required a major change to the PHP engine. Every site using PHP can now safely upgrade their version to stay secure, in addition to applying mitigations from various security vendors.

Due to the severity of these flaws and PHP’s widespread use, we consider this bug to be one of the most impactful vulnerabilities in recent memory. It also marks a new phase, where AI can find security bugs with internet-scale impact.

This is an example of the Defender’s Window in action: to our knowledge, frontier AI was able to find this vulnerability ahead of attackers, giving defenders time to prepare.
https://t.co/TW3jrSbB3f

— OpenAI (@OpenAI) October 21, 2026

PHP is the code that runs behind many sites when you log in, shop or post a comment. WordPress, the software behind millions of blogs and stores, is built on it. A flaw in PHP itself puts all of those sites at risk at once.

The PHP security team's advisory says the two bugs, CVE-2026-109618 and CVE-2026-109619, can let an attacker run their own code on a server without logging in. It rates the risk highest for Apache servers using mod_php, where a single request can trigger the attack before any site code runs. Any PHP setup that serves web pages is affected.

OpenAI says security firm Calif found one flaw using its GPT-5.6 Sol model through Patch the Planet, an OpenAI program that pairs AI bug hunting with human reviewers. Cloud security company Wiz also found and reported both flaws on its own, according to OpenAI. PHP's volunteer maintainers wrote the fixes. OpenAI says it spent two months helping coordinate the response with partners such as Cloudflare and US government agencies, including CISA.

The plan did not go perfectly. Details leaked a day early. BleepingComputer reported that security firm watchTowr confirmed the flaws were real after the leak and said it had reproduced one. PHP warned admins in advance and shipped the fixes the next day. The EU's cyber response team warned organisations to prepare to patch at once.

The bigger story is what AI now does in security. OpenAI calls this a Defender's Window, where AI finds a bug before criminals do. That holds only while defenders have better tools than attackers. OpenAI admits the same skills can help attackers build exploits, and it expects findings like this to rise sharply in coming months. That means more urgent patches for the small volunteer teams who maintain open-source code.

OpenAI is putting money behind its pitch. It says it has committed $1 billion in subsidised access to its security tools for groups with limited resources that protect essential services. It also published a draft playbook for handling future AI-found bugs and is asking the security community for feedback.

Some claims remain unproven. OpenAI says AI found these flaws before attackers, but it qualifies that as "to our knowledge." The 70% figure is OpenAI's estimate. watchTowr said it had seen no attacks in the wild before the patch, but that can change quickly now that the fix shows where the bugs are.

What site owners should do

If you run a PHP site, update now to PHP 8.5.12, 8.4.27, 8.3.36 or 8.2.35, or install your Linux distribution's patched package. PHP 8.1 and older no longer get official updates, so check with your host or upgrade. If you use managed hosting or a WordPress host, ask your provider whether it has patched. Regular visitors to websites don't need to do anything.