AI models find decade-old PHP vulnerabilities
Two critical security flaws in PHP, which powers roughly 70% of websites, were identified and patched after remaining undetected for over ten years. The discovery highlights the growing role of artificial intelligence in uncovering complex code issues that evade traditional human auditing.
Two critical security vulnerabilities in the PHP programming language have been discovered and patched, marking a significant shift in how long-standing code defects are identified. These flaws had existed for more than a decade without being detected by human researchers. The vulnerabilities were identified with the assistance of AI models from OpenAI and Google, which helped both locate and name the specific security gaps. PHP is a foundational technology for the web, powering approximately 70% of all websites, which makes the discovery of such persistent flaws particularly consequential for digital infrastructure. The application of these patches resulted in temporary unavailability for some websites as system administrators worked to implement the necessary fixes. This incident serves as a concrete example of the utility of AI in cybersecurity, demonstrating its capacity to detect complex, long-standing code issues that traditional human auditing methods have failed to catch. The successful identification and closure of these gaps underscore the potential for machine learning tools to enhance the security posture of widely used software platforms.