How did AI discover the internet's massive cyber-security risk?

As of October 21, 2026, critical vulnerabilities were discovered in PHP, a widely used web programming language, prompting European and British authorities to recommend that many websites temporarily go dark. This led to several major sites, including government pages in Europe, the BBC, and other organizations like Wikipedia and Pinboard, being taken offline on Wednesday morning as website operators scrambled to apply security updates or risk potential cyberattacks. The vulnerabilities, rated at a critical severity of 9.2 out of 10, could allow hackers to remotely execute code, potentially compromising sensitive data and networks. Reports suggest that up to 70% of the world's websites may be affected by these issues, as PHP underpins large parts of the internet infrastructure. The discovery underscores concerns about the rapid spread of AI in cybersecurity, with some experts calling it "one of the first areas where we are going to see the harmful impact of AI."