AI-driven security flaws expose global internet vulnerabilities

A wave of cyber-security incidents has disrupted websites globally, forcing security teams into emergency action after critical vulnerabilities were exposed in widely used software and encryption systems. The turmoil began on October 20th when major web browsers received an email containing eight insecure private security keys, some as short as a single digit, originally issued by an Indian authority. The following day, a severe vulnerability in the PHP programming language was revealed, prompting governments to advise taking affected systems offline, with several European government websites going dark. Both the PHP flaw and the weak security keys appear to have been discovered by AI, according to officials and researchers. Further vulnerabilities were later found in GitLab and Oracle's Java, highlighting a new challenge: while AI helps uncover weaknesses, publicizing fixes risks tipping off attackers, who can use AI to exploit the flaws before patches are installed.