Black October: An expert explains how AI is changing cybersecurity

Last week, a series of major cybersecurity incidents occurred, leading experts to dub October "Black October." On October 17, a researcher discovered weak private keys used for eMudhra certificates, enabling hackers to impersonate legitimate websites. The following day, the PHP Foundation announced two critical-level vulnerabilities, dubbed PHP2Shell, which exploited memory safety issues and could allow remote users to control PHP servers.

Additionally, an open-source repository of LLM providers was hacked and quarantined, all within hours. GitLab also released a fix for a flaw of the highest severity, which could allow hackers to remotely control systems. The key factor linking these incidents was artificial intelligence (AI), which played a role in vulnerability discovery, amplification, and limited support infrastructure.

AI-assisted discovery and ramifications in PHP2Shell, PHP2Shell was discovered using AI-assisted techniques, despite the possibility of such exploits being raised in 2004 and not being able to reproduce a working attack previously. If AI could overcome the obstacles, it could help us show that systems behave as expected or find flaws.

However, the requirement for mathematical expertise and scaling issues hindered its widespread deployment. AI enthusiasts predict that things will get worse before they get better, emphasizing the need for security leaders to ask for help and focus on the basics.