Cloudflare CEO Matthew Prince warns that artificial intelligence is accelerating the discovery of critical software vulnerabilities. The recent PHP security flaw, which affected approximately 70% of websites globally, is considered a precursor to a prolonged period of digital instability. Government sites and platforms like GOG were forced to deploy emergency patches or shut down operations.
Prince revealed that Cloudflare has identified multiple undisclosed critical vulnerabilities in other essential software. These flaws are comparable in severity to the recent PHP issue. The company anticipates that such security crises may occur on a weekly basis in the coming years, keeping the entire industry under constant pressure to respond to new threats.
The shift is driven by the rapid development of large language models. Previously, finding exploitable vulnerabilities required weeks of expert analysis. Now, AI agents can analyze codebases in hours. The security team Wiz Research confirmed that advanced AI models were instrumental in identifying the PHP vulnerability and developing proof-of-concept exploits.
Data from the Electronic Frontier Foundation shows that 3% of all known software vulnerabilities were discovered in October alone. Research by VulnCheck indicates a significant increase in vulnerability detection rates assisted by AI. Experts suggest that while this creates short-term repair burdens, it ultimately exposes long-standing hidden flaws, potentially strengthening internet infrastructure in the long run.
