Jason Clinton, a security executive at Anthropic, warned that AI is detecting software vulnerabilities faster than humans can fix them. This poses a significant threat to critical infrastructure, particularly public services such as water and electricity.
Clinton advised managers of critical infrastructure to isolate sensitive networks from external access to prevent unauthorized access. He also urged security officials to convince their leadership to invest more in cybersecurity.
A recent example of the risk was illustrated when PHP developers released emergency patches for two severe vulnerabilities. These flaws, rated 9.2 out of 10 in severity, could have allowed intruders to execute malicious code on servers without needing a password.
Clinton’s warning comes as AI tools are being used to detect vulnerabilities in systems more quickly than administrators can remediate them. Anthropic has openly discussed using its own AI technology to uncover thousands of bugs in open-source software.
This capability highlights the dual nature of AI, which can both secure systems and rapidly uncover vulnerabilities in digital infrastructure. Clinton recommended that organizations focus on fundamental security practices rather than relying solely on AI.
Read Also: Sweden turns Gotland into NATO’s Baltic air defense hub
He also suggested that security officials take advantage of the current urgency to secure necessary funding and support from their boards. According to Clinton, the current situation may mark the beginning of a prolonged period of instability, with stability potentially not returning for about a year and a half.
Clinton emphasized that it is essential for organizations to prioritize cybersecurity investments to mitigate the risks associated with AI-detected vulnerabilities. They must also ensure that their security practices are robust and can withstand the increasing threat environment.
Anthropic’s experience with AI technology has shown that it can be a powerful tool for identifying vulnerabilities, but it is not a replacement for traditional security measures. The company’s security experts, including Clinton, have stressed the importance of a multi-layered approach to cybersecurity.
As the use of AI tools becomes more widespread, the potential for vulnerabilities to be exploited increases. Clinton warned that organizations must be proactive in addressing these risks and investing in cybersecurity measures to protect their critical infrastructure.