AI Discovery Exposes PHP2Shell Vulnerability Impacting 70% of Global Websites

POLICY WIRE — California, United States — Cybersecurity researchers leveraging an advanced OpenAI artificial intelligence model successfully uncovered a decade-old internet vulnerability before malicious hackers could exploit it.

The security flaw targets a programming language known as PHP, an acronym for Personal Home Page, which currently powers up to 70% of the world’s websites and web applications, including platforms utilized by government agencies and social media giants like Facebook.

This major discovery highlights an ongoing arms race between cybersecurity defenders and malicious actors regarding the use of AI to detect potential technology infrastructure weaknesses. In this instance, security professionals operating in California dubbed the newly found flaw PHP2Shell, utilizing sophisticated AI models that proved essential for identifying the obscure defect.

People understood that something unusual was happening, but nobody could figure out how to write an exploit that escalated a bug causing computer crashes into a critical flaw allowing unauthorized software execution, as noted during the investigation.

Researchers explained that the issue was exceptionally complicated and tedious to uncover, noting that human experts would typically require months of tedious effort to diagnose the underlying mechanics. Conversely, artificial intelligence handles tedious, complex, and long-term analytical tasks at a much faster pace.

Experts emphasized that the time elapsed between vulnerability detection and patch deployment remains the defining factor in modern cybersecurity. The entire game centers on the crucial gap between a patch becoming available and attackers exploiting systems before administrators apply updates.

However, artificial intelligence has dramatically compressed this vital time window, enabling cybercriminals to operate with unprecedented speed compared to previous years. Professionals noted that the primary objective is simply urging organizations to apply software patches in a timely manner.

Responding to the discovery, the PHP Foundation announced on Wednesday that software developers working on the language released comprehensive fixes alongside detailed update instructions. The group warned that anyone running PHP must update immediately, noting that announcements on php.net contain technical specifics regarding vulnerabilities, attack vectors, and system-specific update instructions.

The FBI cyber threat division acknowledged the vulnerabilities through a post on social media platform X, warning that the flaws expose websites and applications to interference from malicious actors capable of remotely injecting malware into PHP source code. The agency cautioned that this exposure could potentially compromise sensitive data and networks.

Consequently, the FBI strongly advises any organization unable to implement immediate patches to take vulnerable web properties offline temporarily. Regular consumers risk having personal data exposed if they maintain accounts with online retailers or booking platforms that fail to secure their infrastructure.

For standard internet users, simply navigating to a PHP-based website does not automatically compromise personal devices. Instead, the primary danger lies in attackers exploiting web platforms storing user data, allowing malicious actors to harvest stored details such as email addresses, physical addresses, names, and dates of birth.

Nevertheless, experts clarified that a site’s vulnerability does not guarantee a security breach has transpired or will transpire. Users have no reason to halt online browsing or shopping entirely, though they should remain highly vigilant regarding suspicious communications because the exact scope of data stolen from specific websites remains unknown.

Reporting by Policy-Wire (PW)