In brief
An AI system, powered by OpenAI’s language model, identified a deep‑seated flaw in PHP that underpins 70 % of web sites. The vulnerability could let attackers hijack sessions and inject malicious code, prompting a rapid patch rollout. Cyber‑security firm Wiz called it the most severe flaw ever detected in core web infrastructure. The incident highlights AI’s potential as a rapid‑response tool for code analysis.
An advanced AI system revealed a staggering vulnerability in PHP, the server‑side language that powers more than 70 % of active websites worldwide. The flaw could let attackers hijack site sessions, exfiltrate data and inject malicious code.
The defect had lurked unnoticed for years, slipping through conventional code audits. With billions of online services relying on the platform, the potential damage spanned e‑commerce, government portals and popular knowledge bases.
Software maintainers rushed to ship patches, issuing a series of updates that addressed a range of out‑of‑bounds reads, hard‑coded defaults and inadequate input validation. Security analysts at Wiz dubbed the issue the most severe flaw yet detected in widely used web infrastructure.
The discovery was made possible by a Natural‑Language‑Processing model built on OpenAI’s GPT‑4 architecture, which systematically reviewed PHP’s codebase, identified anomalies and translated them into actionable bug reports. The event underscores how artificial intelligence can act as a rapid‑response unit for complex security vetting.
Web owners should verify that their hosting providers have applied the latest PHP patches, enable Web Application Firewalls and conduct routine penetration testing. While the fix rolls out, vigilance remains critical to thwart any opportunistic exploitation during the transition.
