OpenAI AI models help researchers uncover critical security flaws in PHP

The company added that the findings signal a new phase in cybersecurity, where frontier AI models can help detect flaws.OpenAI’s artificial intelligence models have helped researchers identify critical security vulnerabilities in PHP, an open-source programming language used to power an estimated 70 per cent of websites worldwide.The discovery triggered a coordinated security effort over the past two months, bringing together software maintainers, researchers and security providers to develop fixes and protective measures before the flaws were publicly disclosed.According to OpenAI, addressing the vulnerabilities required substantial changes to the PHP engine, making the remediation process particularly complex.

With fixes now available, websites running PHP can upgrade to secure versions of the software and apply additional mitigations released by 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,” OpenAI said in a post on X.The development also highlights the growing role of advanced AI models in identifying software vulnerabilities that could have far-reaching consequences for internet security.OpenAI said the discovery demonstrates its concept of the “Defender’s Window”, in which AI-powered security research identifies vulnerabilities before malicious actors can exploit them, giving developers and security teams time to prepare and deploy protections.The company added that the findings signal a new phase in cybersecurity, where frontier AI models can help detect flaws with potentially internet-wide implications and strengthen defences before attackers can take advantage of them.