A senior security manager at Anthropic has warned of the growing danger of cyberattacks, particularly against vital public infrastructure. Jason Clinton, the company’s deputy chief information security officer, warned that a huge wave of software flaws could soon overwhelm existing security systems.
He noted that the integration of artificial intelligence is accelerating the discovery of security vulnerabilities, leaving operators of essential utilities – such as water and electricity – struggling to apply patches in time.
To mitigate the risks, Clinton advised organisations to block external access to their most sensitive networks. This can help them prevent unauthorised access. In a post on X, he suggested that the current situation could mark the beginning of a longer period of deterioration. In his view, stability may not return for around a year and a half.
Clinton also urged cyber security leaders to use the current sense of urgency to secure essential funding and support from their boards. He advocated a return to fundamental security practices, rather than focusing exclusively on AI.
Vulnerabilities emerge more quickly
This warning follows widespread concern that AI-powered tools are enabling hackers to identify system weaknesses faster than administrators can fix them. This risk was recently illustrated when the developers of PHP, a foundational programming language for the Web, rushed out emergency patches for two serious vulnerabilities.
These vulnerabilities, which were given a severity score of 9.2 out of 10, could potentially allow intruders to execute malicious code on servers without needing passwords.
Although Clinton did not specify exactly which events prompted his warning, Anthropic has spoken openly about using its own AI technology to uncover thousands of bugs in open source software.
That capability demonstrates the two sides of AI. The technology can help improve system security, but it can also expose vulnerabilities in digital infrastructure more quickly.
Business AM
Ltd
