Depending on the side you find yourself on, artificial intelligence (AI) is emerging as a powerful tool, without doubt.
If you with the good guys, the tech is one of cybersecurity’s most powerful allies.
But a new global security crisis shows how the same technology that helps researchers uncover hidden software vulnerabilities can also help hackers exploit them.
Which essentially means that, for hackers, AI is becoming a powerful tool for identifying and exploiting security gaps in websites.
On October 21, 2026, many government websites and public services across Europe and elsewhere went offline as organizations scrambled to patch two critical security flaws in PHP, a programming language that powers nearly 70% of websites whose underlying technology can be identified.
Germany, France, Belgium, the Netherlands, Finland and the Czech Republic were among the countries affected. Canada and Australia also reported disruptions to government websites. The scale of the response underlined how a vulnerability in one widely used technology can ripple across the Internet.
The good news: AI can find what humans miss
Security researchers are increasingly using AI to analyze enormous codebases, trace complex program behavior and identify vulnerabilities that conventional reviews might overlook. In this latest episode, researchers associated with Calif and Google-owned cybersecurity company Wiz used AI-assisted techniques to help uncover critical PHP flaws.
The discovery demonstrates how AI can accelerate security research and help developers fix dangerous weaknesses before they become widespread disasters.
PHP developers released security updates, while cybersecurity agencies urged organizations to patch affected systems immediately or take vulnerable websites offline until they could be secured. In this context, a temporary shutdown is not necessarily evidence of a successful attack. It can be a deliberate defensive measure to protect sensitive information and essential services.
The bad news: hackers use AI, too
The same capabilities that help defenders understand complex code can help attackers analyze vulnerabilities, adapt exploit techniques and identify systems that remain unpatched.
Reuters reported signs of early exploitation attempts, adding urgency to the global response. Once a vulnerability becomes public, organizations face a race against time: security fixes may be available, but thousands of administrators still need to install and verify them.
AI does not automatically make the Internet safer or more dangerous. It accelerates what both sides can do. And because PHP is embedded in countless websites, the consequences of a single serious flaw can extend far beyond the original software.
The lesson: finding the bug is only the beginning
Organizations must patch promptly, monitor suspicious activity, maintain reliable backups and isolate vulnerable services when necessary. AI-assisted security testing should become a regular part of software development, not merely an emergency response.
This episode offers a glimpse of cybersecurity’s future: AI helping expose weaknesses that have survived for years, while attackers race to exploit them. The decisive advantage will belong to those who turn discovery into protection before the window for defence closes.