AI turns 22-year-old PHP vulnerability into a two-day crisis

Artificial intelligence exposed a critical vulnerability in PHP, the programming language powering 70% of websites, turning a 22-year-old known threat into a crisis in just two days. OpenAI revealed the exploit Wednesday alongside a security patch. The vulnerability, first warned about in 2004, allowed attackers to steal passwords, access databases, or take full control of websites. Exploiting it manually was so tedious that it remained impractical for humans, but AI found attack methods within days in August. Experts warn this democratization of AI hacking tools signals a broader threat, as many similar vulnerabilities remain unpatched. The patch may not cover all PHP versions, and thousands of attack attempts occurred before the official announcement.